From jomegat at jomegat.com Fri Sep 2 03:09:54 2011 From: jomegat at jomegat.com (Jomegat) Date: Thu, 01 Sep 2011 23:09:54 -0400 Subject: [K12OSN] Local Apps and Firefox 6 In-Reply-To: <4E5AE383.7070505@jomegat.com> References: <4E5A5ABC.9020204@jomegat.com> <4E5A8900.7060709@jomegat.com> <4E5AE383.7070505@jomegat.com> Message-ID: <4E604902.2020109@jomegat.com> I have abandoned local apps. I got them working, but flash was slower when running locally than it was when running on the server. Adobe says you need 512M to run flash 10, and my clients have only 256M. Then there were several other plugins (to run quicktime for example) that had not yet been installed as local apps, so I threw in the towel there. Back to running everything on the server. As per advice received here, I reconfigured my managed switch to reduce the timeout values in the Spanning Tree. Once that was done, the clients were able to boot through the switch, so it is now sitting between my TS & TC's. Nice. And now that I have a gigabit pipe to the server, flash performance running on the server is acceptable. I'll find out tomorrow if there are any other gotcha's. On 08/28/2011 08:55 PM, Jomegat wrote: > Thank you Robert. Your advice was absolutely necessary, but I also > needed to copy /etc/resolv.conf to /opt/ltsp/i386/etc - that solved the > yum problem. > > I also found that ltsp-server-tweaks has to be run when logged into the > server at the console. Running it while remotely logged in via ssh was > not enough? (Irene abated sufficiently for me to drive to the school and > fool around with the network.) > > I now have firefox (and flash, and tuxpaint) installed as local apps, > and they seem to run OK. Now all I need to do is figure out how to apply > that change to all the user accounts. I don't want to log into each one > and go through gnome's "add launcher" dialog. There must be a way to > script this. > > I made a stab by copying the files in my own > ~/.gnome2/panel2.d/defaults/launcher to that same directory in everyone > else's acct. I also chowned it and chmodded it - but for whatever > reason, the applets don't show up on the menu. Does anyone know how to > change the default applets? Right now I get firefox, evolution, and > gnotes in the menu panel at the top. I'd much rather have > localapp-firefox, tuxpaint, and OpenOffice up there. > > I will do some research while I wait for an answer. I don't use gnome > much myself, but rather, have been using KDE since... forever. > > Thanks again for the help. You guys have been golden. > > On 08/28/2011 02:29 PM, Jomegat wrote: >> On 08/28/2011 02:10 PM, Robert Arkiletian wrote: >>> On Sun, Aug 28, 2011 at 8:11 AM, Jomegat wrote: >>>> Hi everyone, >>>> >>>> I'm planning to run Firefox as a local application and I'm looking >>>> for some >>>> advice. First I tried following the instructions at >>>> https://fedorahosted.org/k12linux/wiki/LocalAppsSupport >>> >>> Not 100% sure as I can't verify this. It's been a while, but check if >>> /proc is mounted inside the chroot. I think you used to have to >>> >>> mount /proc >>> >>> inside the chroot before you do yum install. >> >> If that helped, it didn't help enough: >> >> [root at termserv ~]# cd /opt/ltsp/i386/ >> [root at termserv i386]# setarch i386 chroot . >> bash-4.1# mount /proc >> bash-4.1# yum install firefox >> Error: Cannot retrieve repository metadata (repomd.xml) for repository: >> epel. Please verify its path and try again >> bash-4.1# >> >>> >>>> >>>> but when I got to the yum part, there was no joy: >>>> >>>> [root at termserv i386]# setarch i386 chroot . >>>> bash-4.1# yum install firefox >>>> Error: Cannot retrieve repository metadata (repomd.xml) for repository: >>>> epel. Please verify its path and try again >>>> >>>> I thought - maybe I need to run ltsp-server-tweaks, since I do not >>>> recall >>>> having run that when I last rebooted (but everything seemed to work >>>> then, >>>> so... I dunno?) >>>> >>>> Here's what happens when I try running that: >>>> [root at termserv ~]# ltsp-server-tweaks >>>> Failed to access configuration source(s): Failed to contact >>>> configuration >>>> server; some possible causes are that you need to enable TCP/IP >>>> networking >>>> for ORBit, or you have stale NFS locks due to a system crash. See >>>> http://projects.gnome.org/gconf/ for information. (Details - 1: Not >>>> running >>>> within active session) >>>> >>>> Hmmm. Not sure what all that means. >>>> >>>> I tried all this while remotely logged in, as Tropical Storm Irene >>>> might >>>> make it a bit dodgy to get to the school (or get home afterwards). I >>>> don't >>>> think that's significant, but just in case it is, I'm disclosing it >>>> here. >>>> >>>> My next thought was that maybe I should try FF 6 instead of 3.6 >>>> (which is >>>> what yum will install). Is there a reason I should avoid that? >>>> >>>> Thanks in advance! >>>> >>>> -- >>>> Jim Thomas jomegat at jomegat.com >>>> Dyslexics of the world, UNTIE! >>>> >>>> _______________________________________________ >>>> K12OSN mailing list >>>> K12OSN at redhat.com >>>> https://www.redhat.com/mailman/listinfo/k12osn >>>> For more info see >>>> >>> >>> >>> >> >> > > -- Jim Thomas jomegat at jomegat.com Getting an inch of snow is like winning ten cents in the lottery - Calvin From warren at togami.com Sat Sep 3 20:31:32 2011 From: warren at togami.com (Warren Togami Jr.) Date: Sat, 03 Sep 2011 10:31:32 -1000 Subject: [K12OSN] New EL6 install In-Reply-To: <4E580098.1030209@jomegat.com> References: <4E57E9EA.6010909@jomegat.com> <4E57FE76.6090806@astrofoto.org> <4E580098.1030209@jomegat.com> Message-ID: <4E628EA4.3000705@togami.com> On 8/26/2011 10:22 AM, Jomegat wrote: > I had no trouble with youtube either. But there was another flash-heavy > site that had real problems. I'm not sure what site the students were > using, but it brought everything to its knees whatever it was. > > Flash video uses a LOT of bandwidth. That is likely killing your 100mbit switch. Gigabit switch will improve it a bit, but your only choice is to switch to local apps if you want flash video on all clients simultaneously to not kill the network. Warren From bfristen at shaw.ca Sun Sep 4 21:21:12 2011 From: bfristen at shaw.ca (Brian Fristensky) Date: Sun, 04 Sep 2011 16:21:12 -0500 Subject: [K12OSN] Frustration with RHEL6.1 Message-ID: <4E63EBC8.9010708@shaw.ca> I just did a clean install of RHEL6.1 on my server. I did the standard Desktop system install on an AMD64 workstation. Is it only me, but has anyone else had minimal luck installing anything on this platform? It seams like whenever I try to install software using yum or the GNOME graphical installer, I get endless dependency errors. On recent versions of Fedora, yum seemed to always be able to find and install dependencies automatically. Right out of the box. This is all relevant to LTSP. I added to following file to /etc/yum.repos.d [LTSP] name=Togami LTSP Testing Repo baseurl=http://mplug.org/~k12linux/rpm/el6/$basearch/ enabled=1 gpgcheck=0 gpgkey=http://www.dageek.co.uk/ltsp/RPM-GPG-KEY-DaGeek-LTSP and then tried yum install ltsp-server: ... skipping the list of packages to install... --> Finished Dependency Resolution Error: Package: 1:python-imgcreate-13.3-1.el6.x86_64 (epel) Requires: pykickstart >= 0.96 Error: Package: ltsp-server-5.2.17-1.el6.x86_64 (LTSP) Requires: perl(NetAddr::IP) Error: Package: 1:livecd-tools-13.3-1.el6.x86_64 (epel) Requires: syslinux Error: Package: 1:python-imgcreate-13.3-1.el6.x86_64 (epel) Requires: squashfs-tools Error: Package: 1:livecd-tools-13.3-1.el6.x86_64 (epel) Requires: pyparted Error: Package: 1:livecd-tools-13.3-1.el6.x86_64 (epel) Requires: isomd5sum Error: Package: ltsp-server-5.2.17-1.el6.x86_64 (LTSP) Requires: dhcp Error: Package: 1:livecd-tools-13.3-1.el6.x86_64 (epel) Requires: /sbin/extlinux Error: Package: ltsp-server-5.2.17-1.el6.x86_64 (LTSP) Requires: tftp-server You could try using --skip-broken to work around the problem You could try running: rpm -Va --nofiles --nodigest As I've said, it's not just ltsp. It seems that the majority of the time, when I try to install packages with yum, it fails to resolve dependency errors. I have lots of repos in which yum can find things: -rwxr-xr-x. 1 root root 808 Jan 25 2009 atrpms-bleeding.repo -rwxr-xr-x. 1 root root 748 Jan 25 2009 atrpms.repo -rwxr-xr-x. 1 root root 799 Jan 25 2009 atrpms-testing.repo -rw-r--r--. 1 root root 957 Oct 12 2010 epel.repo -rw-r--r--. 1 root root 1056 Oct 12 2010 epel-testing.repo -rw-r--r--. 1 root root 248 Aug 12 01:32 k12linux.repo -rw-r--r--. 1 root root 172 Jun 21 17:13 ltsp.repo -rw-r--r--. 1 root root 67 Sep 4 15:43 redhat.repo -rw-r--r--. 1 root root 529 Apr 27 12:58 rhel-source.repo -rw-r--r--. 1 root root 1002 May 14 11:46 rpmfusion-free-updates.repo -rw-r--r--. 1 root root 1062 May 14 11:46 rpmfusion-free-updates-testing.repo -rw-r--r--. 1 root root 1047 May 14 11:49 rpmfusion-nonfree-updates.repo -rw-r--r--. 1 root root 1107 May 14 11:49 rpmfusion-nonfree-updates-testing.repo but that doesn't seem to make a difference. One other point. As of RHEL6.1, numerous programs that always used to be present in Fedora are now gone. These include Mozilla seamonkey VNC gFTP uudeview gkrell keychain others look very challenging to install, due to dependencies: nedit tgif ... and many others I have tried so far. I feel like I'm back in the early 2000's, in terms of the frustration with what should be a polished, commercial Linux. Any bright ideas? From jim.kinney at gmail.com Sun Sep 4 22:13:03 2011 From: jim.kinney at gmail.com (Jim Kinney) Date: Sun, 4 Sep 2011 18:13:03 -0400 Subject: [K12OSN] Frustration with RHEL6.1 In-Reply-To: <4E63EBC8.9010708@shaw.ca> References: <4E63EBC8.9010708@shaw.ca> Message-ID: You will need to instal again and select "customize now" on the screen where you selected desktop install. Once done, you have a file in /root called anaconda.log which is a kickstart file for replicating that install. OR... something is hoaed with yum in the repo files or network access to the repos. On Sep 4, 2011 5:27 PM, "Brian Fristensky" wrote: > I just did a clean install of RHEL6.1 on my server. I did the standard Desktop > system > install on an AMD64 workstation. > > Is it only me, but has anyone > else had minimal luck installing anything on this platform? It seams like > whenever I try to install software using yum or the GNOME graphical installer, > I get endless dependency errors. > > On recent versions of Fedora, yum seemed to always be able to find and > install dependencies automatically. Right out of the box. > > This is all relevant to LTSP. I added to following file to /etc/yum.repos.d > > [LTSP] > name=Togami LTSP Testing Repo > baseurl=http://mplug.org/~k12linux/rpm/el6/$basearch/ > enabled=1 > gpgcheck=0 > gpgkey=http://www.dageek.co.uk/ltsp/RPM-GPG-KEY-DaGeek-LTSP > > and then tried yum install ltsp-server: > > ... skipping the list of packages to install... > > --> Finished Dependency Resolution > Error: Package: 1:python-imgcreate-13.3-1.el6.x86_64 (epel) > Requires: pykickstart >= 0.96 > Error: Package: ltsp-server-5.2.17-1.el6.x86_64 (LTSP) > Requires: perl(NetAddr::IP) > Error: Package: 1:livecd-tools-13.3-1.el6.x86_64 (epel) > Requires: syslinux > Error: Package: 1:python-imgcreate-13.3-1.el6.x86_64 (epel) > Requires: squashfs-tools > Error: Package: 1:livecd-tools-13.3-1.el6.x86_64 (epel) > Requires: pyparted > Error: Package: 1:livecd-tools-13.3-1.el6.x86_64 (epel) > Requires: isomd5sum > Error: Package: ltsp-server-5.2.17-1.el6.x86_64 (LTSP) > Requires: dhcp > Error: Package: 1:livecd-tools-13.3-1.el6.x86_64 (epel) > Requires: /sbin/extlinux > Error: Package: ltsp-server-5.2.17-1.el6.x86_64 (LTSP) > Requires: tftp-server > You could try using --skip-broken to work around the problem > You could try running: rpm -Va --nofiles --nodigest > > As I've said, it's not just ltsp. It seems that the majority of the time, when I try > to install packages with yum, it fails to resolve dependency errors. I have lots of > repos in which yum can find things: > > -rwxr-xr-x. 1 root root 808 Jan 25 2009 atrpms-bleeding.repo > -rwxr-xr-x. 1 root root 748 Jan 25 2009 atrpms.repo > -rwxr-xr-x. 1 root root 799 Jan 25 2009 atrpms-testing.repo > -rw-r--r--. 1 root root 957 Oct 12 2010 epel.repo > -rw-r--r--. 1 root root 1056 Oct 12 2010 epel-testing.repo > -rw-r--r--. 1 root root 248 Aug 12 01:32 k12linux.repo > -rw-r--r--. 1 root root 172 Jun 21 17:13 ltsp.repo > -rw-r--r--. 1 root root 67 Sep 4 15:43 redhat.repo > -rw-r--r--. 1 root root 529 Apr 27 12:58 rhel-source.repo > -rw-r--r--. 1 root root 1002 May 14 11:46 rpmfusion-free-updates.repo > -rw-r--r--. 1 root root 1062 May 14 11:46 rpmfusion-free-updates-testing.repo > -rw-r--r--. 1 root root 1047 May 14 11:49 rpmfusion-nonfree-updates.repo > -rw-r--r--. 1 root root 1107 May 14 11:49 rpmfusion-nonfree-updates-testing.repo > > but that doesn't seem to make a difference. > > One other point. As of RHEL6.1, numerous programs that always used to be present > in Fedora are now gone. These include > > Mozilla seamonkey > VNC > gFTP > uudeview > gkrell > keychain > > others look very challenging to install, due to dependencies: > > nedit > tgif > ... and many others I have tried so far. > > I feel like I'm back in the early 2000's, in terms of the frustration with what > should be a polished, commercial Linux. > > Any bright ideas? > > > > > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see -------------- next part -------------- An HTML attachment was scrubbed... URL: From gspurgeon at dageek.co.uk Sun Sep 4 22:20:39 2011 From: gspurgeon at dageek.co.uk (Gavin Spurgeon) Date: Sun, 04 Sep 2011 23:20:39 +0100 Subject: [K12OSN] Frustration with RHEL6.1 In-Reply-To: <4E63EBC8.9010708@shaw.ca> References: <4E63EBC8.9010708@shaw.ca> Message-ID: <4E63F9B7.5030300@dageek.co.uk> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi Brian, > I get endless dependency errors. > > On recent versions of Fedora, yum seemed to always be able to find and > install dependencies automatically. Right out of the box. > I feel like I'm back in the early 2000's, in terms of the frustration > with what > should be a polished, commercial Linux. Really quick and simple question, please don't take this the wrong way, But once the install was complete, did you register you RHEL System on RHN so that it has access to all of the repo's ? Without registration RHEL dose not give you any access to the Official Red Hat Repos. - -- Gavin Spurgeon. AKA Da Geek - ---------------------------------------------------------------------- "The happiest of people don't necessarily have the best of everything, they just make the most of everything that comes along their way.." -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.12 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk5j+bYACgkQvp6arS3vDirvdACgiqQzp+VRR+8JO6of8hkdkd0W kngAnjJG3k1sRVnSPIUxJ4h2P0mvmCcM =NqUm -----END PGP SIGNATURE----- -- This message was scanned by DaGeek Spam Filter and is believed to be clean. From gspurgeon at dageek.co.uk Sun Sep 4 22:23:08 2011 From: gspurgeon at dageek.co.uk (Gavin Spurgeon) Date: Sun, 04 Sep 2011 23:23:08 +0100 Subject: [K12OSN] Frustration with RHEL6.1 In-Reply-To: <4E63EBC8.9010708@shaw.ca> References: <4E63EBC8.9010708@shaw.ca> Message-ID: <4E63FA4C.7010203@dageek.co.uk> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > This is all relevant to LTSP. I added to following file to /etc/yum.repos.d > > [LTSP] > name=Togami LTSP Testing Repo > baseurl=http://mplug.org/~k12linux/rpm/el6/$basearch/ > enabled=1 > gpgcheck=0 > gpgkey=http://www.dageek.co.uk/ltsp/RPM-GPG-KEY-DaGeek-LTSP And with 'gpgcheck=0' in the stanza you do not need to reference My GPG Key, Also Warren's .rpm's will not be signed with My GPG Key, I would assume that Warren would use his own GPG Key, If he has signed his .rpm's... - -- Gavin Spurgeon. AKA Da Geek - ---------------------------------------------------------------------- "The happiest of people don't necessarily have the best of everything, they just make the most of everything that comes along their way.." -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.12 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk5j+kwACgkQvp6arS3vDirhfgCg4kC7WFos4gntBiXbbfboATfX eYgAnRcxDG0SpxD/qn99DX6RbhAGTrS1 =fr7W -----END PGP SIGNATURE----- -- This message was scanned by DaGeek Spam Filter and is believed to be clean. From bfristen at shaw.ca Sun Sep 4 22:31:57 2011 From: bfristen at shaw.ca (Brian Fristensky) Date: Sun, 04 Sep 2011 17:31:57 -0500 Subject: [K12OSN] Frustration with RHEL6.1 In-Reply-To: References: <4E63EBC8.9010708@shaw.ca> Message-ID: <4E63FC5D.90504@shaw.ca> I will make one other observation, which is the very first thing I did after a clean install was to run 'yum update'. Normally, that results in about an hour or so of downloads and installs by yum. This time, after looking at the Red Hat repo, said that the system was up to date. That does sound a bit fishy, since this was an install using the RHEL 6.1 DVD image. Has anybody else actually worked with RHEL 6.1 ? -- ============================================ Brian Fristensky 971 Somerville Avenue Winnipeg MB R3T 1B4 CANADA bfristen at shaw.ca 204-261-3960 ============================================ From gspurgeon at dageek.co.uk Sun Sep 4 22:50:57 2011 From: gspurgeon at dageek.co.uk (Gavin Spurgeon) Date: Sun, 04 Sep 2011 23:50:57 +0100 Subject: [K12OSN] Frustration with RHEL6.1 In-Reply-To: <4E63FC5D.90504@shaw.ca> References: <4E63EBC8.9010708@shaw.ca> <4E63FC5D.90504@shaw.ca> Message-ID: <4E6400D1.2040004@dageek.co.uk> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 04/09/2011 23:31, Brian Fristensky wrote: > I will make one other observation, which is the very first thing I did > after a clean > install was to run 'yum update'. > > Normally, that results in about an hour or so of downloads and installs > by yum. > This time, after looking at the Red Hat repo, said that the system was > up to date. > That does sound a bit fishy, since this was an install using the RHEL > 6.1 DVD > image. > > Has anybody else actually worked with RHEL 6.1 ? Brian, I work for Red Hat. To me, everything you have said so far, points to the fact that you have not registered the system. So you do not have access to any of the repo's. Can you confirm that you have registered the server to RHN ? Do you have a Legitimate Subscription for using RHEL6 (Even an RHN Evaluation Subscription) ? - -- Gavin Spurgeon. AKA Da Geek - ---------------------------------------------------------------------- "The happiest of people don't necessarily have the best of everything, they just make the most of everything that comes along their way.." -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.12 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk5kANEACgkQvp6arS3vDirCyQCgsqisjeK5CD/u9ZGz/Nyz8Wv6 +gsAnipyd15n5L85jG4IpibD5fQ+DpNz =5aJh -----END PGP SIGNATURE----- -- This message was scanned by DaGeek Spam Filter and is believed to be clean. From bfristen at shaw.ca Sun Sep 4 23:11:09 2011 From: bfristen at shaw.ca (Brian Fristensky) Date: Sun, 04 Sep 2011 18:11:09 -0500 Subject: [K12OSN] Frustration with RHEL6.1 In-Reply-To: <4E63F9B7.5030300@dageek.co.uk> References: <4E63EBC8.9010708@shaw.ca> <4E63F9B7.5030300@dageek.co.uk> Message-ID: <4E64058D.4030004@shaw.ca> No I haven't registered, at least not as part of the install. System --> RHN Registration has no result. Nothing pops up. The same is true of System --> Red Hat Subscription Manager I have an acknowledgement of my order saying "Your Red Hat subscription is now active". It also gives me an acct. # and Contract #. When I received the acknowledgement, I was also able to log into my acct. at redhat.com. So that end should be okay. Gavin Spurgeon wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > > Hi Brian, > >> I get endless dependency errors. >> >> On recent versions of Fedora, yum seemed to always be able to find and >> install dependencies automatically. Right out of the box. > > >> I feel like I'm back in the early 2000's, in terms of the frustration >> with what >> should be a polished, commercial Linux. > Really quick and simple question, please don't take this the wrong way, > But once the install was complete, did you register you RHEL System on > RHN so that it has access to all of the repo's ? > > Without registration RHEL dose not give you any access to the Official > Red Hat Repos. > > - -- > > Gavin Spurgeon. > AKA Da Geek > > - ---------------------------------------------------------------------- > "The happiest of people don't necessarily have the best of everything, > they just make the most of everything that comes along their way.." > -----BEGIN PGP SIGNATURE----- > Version: GnuPG/MacGPG2 v2.0.12 (Darwin) > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ > > iEYEARECAAYFAk5j+bYACgkQvp6arS3vDirvdACgiqQzp+VRR+8JO6of8hkdkd0W > kngAnjJG3k1sRVnSPIUxJ4h2P0mvmCcM > =NqUm > -----END PGP SIGNATURE----- > > -- > This message was scanned by DaGeek Spam Filter and is believed to be clean. > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > -- ============================================ Brian Fristensky 971 Somerville Avenue Winnipeg MB R3T 1B4 CANADA bfristen at shaw.ca 204-261-3960 ============================================ From gspurgeon at dageek.co.uk Sun Sep 4 23:28:54 2011 From: gspurgeon at dageek.co.uk (Gavin Spurgeon) Date: Mon, 05 Sep 2011 00:28:54 +0100 Subject: [K12OSN] Frustration with RHEL6.1 In-Reply-To: <4E64058D.4030004@shaw.ca> References: <4E63EBC8.9010708@shaw.ca> <4E63F9B7.5030300@dageek.co.uk> <4E64058D.4030004@shaw.ca> Message-ID: <4E6409B6.8090604@dageek.co.uk> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 05/09/2011 00:11, Brian Fristensky wrote: > No I haven't registered, at least not as part of the install. > > System --> RHN Registration > > has no result. Nothing pops up. > > The same is true of > > System --> Red Hat Subscription Manager > > I have an acknowledgement of my order saying "Your Red Hat subscription > is now active". > It also gives me an acct. # and Contract #. > > When I received the acknowledgement, I was also able to log into my > acct. at > redhat.com. So that end should be okay. You still have to register the install to your account before the yum repo settings are activated on your server. You could try to open a root shell and type the command rhn_register --nox And then follow the instructions, it will ask you for your RHN Username and Password and once it has finished, the repo config will be added to you and then you can do a yum update and so on without any issues. - -- Gavin Spurgeon. AKA Da Geek - ---------------------------------------------------------------------- "The happiest of people don't necessarily have the best of everything, they just make the most of everything that comes along their way.." -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.12 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk5kCbUACgkQvp6arS3vDirNQACg0qIKHTI5+iRQhVp1iAZKAMEJ kDMAn3DDYJelVlcYqkoyccqc3Qb6w5GK =QRoQ -----END PGP SIGNATURE----- -- This message was scanned by DaGeek Spam Filter and is believed to be clean. From carl at snarlnet.com Mon Sep 5 19:00:06 2011 From: carl at snarlnet.com (Carl Keil) Date: Mon, 05 Sep 2011 12:00:06 -0700 Subject: [K12OSN] Frustration with RHEL6.1 (Jim Kinney) Message-ID: <4E651C36.3000807@snarlnet.com> Maybe my experience has nothing to do with your problem, but I had very, very similar yum symptoms after my recent install of Centos 5.6. It turned out that my router, for some reason, wasn't routing to my server. I think it was because I was using the same server, and IP address as the old install, but I swapped out a different NIC. (I was getting back on my feet after the death of my boot drive.) My theory is that somehow the gateway to the internet was remembering the old MAC address and not simply relaying things to the old IP. As soon as I undid the port forwarding on the router and reset it, everything was like the good old days of 2009, not 2000. I know what you mean, I had that exact same thought during my ordeal. It was a really bad, frustrating flashback. Both the feeling that Linux wasn't "mature" and that I didn't know crap about how to use it. In the end, it had nothing to do with Linux. Try starting a yum updating and then running "netstat" (I'm pretty sure that's what it was?) Or the gui "System Monitor". Those seemed to throw my NIC into "promiscuous" mode and all of a sudden things would download. Could any of that be relevant to your situation? Have you rebooted/reset all the network equipment between your server and the world? HTH ck From bfristen at shaw.ca Mon Sep 5 23:39:05 2011 From: bfristen at shaw.ca (Brian Fristensky) Date: Mon, 05 Sep 2011 18:39:05 -0500 Subject: [K12OSN] Frustration with RHEL6.1 (Jim Kinney) In-Reply-To: <4E651C36.3000807@snarlnet.com> References: <4E651C36.3000807@snarlnet.com> Message-ID: <4E655D99.3030300@shaw.ca> We'll see. My gut feeling is that Gavin is right, and that it's because I didn't register my Red Hat license yet. Which I would have done during the install. Which I couldn't do right away because RedHat isn't recognizing my password, which did work a week ago, and the automated 'lost password' mechanism isn't responding to me, and this being Labor Day weekend, neither, so far, has customer service. I will make a point of posting when I get this sorted out for others to benefit. But in particular, the observation that yum simply refused to acknowledge that there was anything to update on a new install sounds seems compatible with the presumption that if you don't register, updates are blocked. I should be able to test that as soon as I get a working Red Hat password. Carl Keil wrote: > Maybe my experience has nothing to do with your problem, but I had very, very > similar yum symptoms after my recent install of Centos 5.6. It turned out > that my router, for some reason, wasn't routing to my server. I think it was > because I was using the same server, and IP address as the old install, but I > swapped out a different NIC. (I was getting back on my feet after the death > of my boot drive.) My theory is that somehow the gateway to the internet was > remembering the old MAC address and not simply relaying things to the old > IP. As soon as I undid the port forwarding on the router and reset it, > everything was like the good old days of 2009, not 2000. I know what you > mean, I had that exact same thought during my ordeal. It was a really bad, > frustrating flashback. Both the feeling that Linux wasn't "mature" and that I > didn't know crap about how to use it. In the end, it had nothing to do with > Linux. > > Try starting a yum updating and then running "netstat" (I'm pretty sure that's > what it was?) Or the gui "System Monitor". Those seemed to throw my NIC into > "promiscuous" mode and all of a sudden things would download. > > Could any of that be relevant to your situation? Have you rebooted/reset all > the network equipment between your server and the world? > > HTH > > ck > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > -- ============================================ Brian Fristensky 971 Somerville Avenue Winnipeg MB R3T 1B4 CANADA bfristen at shaw.ca 204-261-3960 ============================================ From roland at astrofoto.org Wed Sep 7 15:27:13 2011 From: roland at astrofoto.org (Roland Roberts) Date: Wed, 07 Sep 2011 11:27:13 -0400 Subject: [K12OSN] drm i915_hangcheck_elapsed on client Message-ID: <4E678D51.4020807@astrofoto.org> I've had my client freeze up, temporarily, a number of times. Not every day, not with high-frequency, but more than once a week. Today, after it recovered, I launched a local xterm and found this in /var/log/messages Sep 7 11:05:08 client-192 kernel: [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU idle, missed IRQ. The previous log message is from the previous evening, the next if from me lauching the local xterm. The client is a Thinkpad 420s with a Core i5 and the integrated Intel GPU. Since that's a new addition for RHEL/SL 6.1, it's probably not client related, but I thought I'd bounce it around here first. roland -- PGP Key ID: 66 BC 3B CD Roland B. Roberts, PhD RL Enterprises roland at rlenter.com 6818 Madeline Court roland at astrofoto.org Brooklyn, NY 11220 From ltsp at symbio-technologies.com Wed Sep 7 16:20:41 2011 From: ltsp at symbio-technologies.com (Gideon Romm) Date: Wed, 7 Sep 2011 12:20:41 -0400 Subject: [K12OSN] Thin client parts on offer Message-ID: Hello, all. I know for many of you, September marks the end of summer and beginning of a new year, and though Spring is usually for cleaning, we have chosen to make some room now. So, I have a partial list of odds and ends that may be of interest to someone on this list. I know many folks here may look for a way to get a few more thin clients at a reduced price. Consider the prices on this list a suggested price. We would like to get all this stuff cleared out by the end of the month, so make us an offer. Prices obviously don't include shipping and I am hoping they will find a home in North America (we are in New York), so as not to have to worry about the various duties associated with selling abroad. Here is a link to the list: http://symbio-technologies.com/LTSP-Sale-201109-total.pdf As always, you all know where to find me. Happy New Year! :) -Gadi -- Gideon Romm Symbio Technologies | www.symbio-technologies.com (o) 914.576.1205 | (f) 914.576.0944 From bfristen at shaw.ca Thu Sep 8 00:57:05 2011 From: bfristen at shaw.ca (Brian Fristensky) Date: Wed, 07 Sep 2011 19:57:05 -0500 Subject: [K12OSN] Frustration with RHEL6.1 (Jim Kinney) In-Reply-To: <4E655D99.3030300@shaw.ca> References: <4E651C36.3000807@snarlnet.com> <4E655D99.3030300@shaw.ca> Message-ID: <4E6812E1.8050601@shaw.ca> Yep. The reason I couldn't install anything turned out to be that I hadn't yet registered my copy of RHEL with RedHat. One interesting thing to come of this is the REASON I couldn't register. I am not certain yet, but it appears that the login at RedHat.com is kind of flakey with certain browsers. I have tried SeaMonkey on two different machines, and in both cases, if you type in the correct userid and password, you get kicked back out to the login prompt. I was successful with Google Chrome, and had mixed luck with Firefox on two different machines. I need to do a few more experiments. For example, with SeaMonkey, even deleting all cookies didn't help. I need to distinguish between the browser per se, and some setting of the browser. Brian Fristensky wrote: > We'll see. My gut feeling is that Gavin is right, and that it's because I > didn't register > my Red Hat license yet. Which I would have done during the install. Which I > couldn't do right > away because RedHat isn't recognizing my password, which did work a week ago, and > the automated 'lost password' mechanism isn't responding to me, and this being > Labor > Day weekend, neither, so far, has customer service. > > I will make a point of posting when I get this sorted out for others to benefit. > But in particular, the observation that yum simply refused to acknowledge > that there was anything to update on a new install sounds seems compatible > with the presumption that if you don't register, updates are blocked. I should > be able to test that as soon as I get a working Red Hat password. > > Carl Keil wrote: >> Maybe my experience has nothing to do with your problem, but I had very, very >> similar yum symptoms after my recent install of Centos 5.6. It turned out >> that my router, for some reason, wasn't routing to my server. I think it was >> because I was using the same server, and IP address as the old install, but I >> swapped out a different NIC. (I was getting back on my feet after the death >> of my boot drive.) My theory is that somehow the gateway to the internet was >> remembering the old MAC address and not simply relaying things to the old >> IP. As soon as I undid the port forwarding on the router and reset it, >> everything was like the good old days of 2009, not 2000. I know what you >> mean, I had that exact same thought during my ordeal. It was a really bad, >> frustrating flashback. Both the feeling that Linux wasn't "mature" and that >> I didn't know crap about how to use it. In the end, it had nothing to do >> with Linux. >> >> Try starting a yum updating and then running "netstat" (I'm pretty sure >> that's what it was?) Or the gui "System Monitor". Those seemed to throw my >> NIC into "promiscuous" mode and all of a sudden things would download. >> >> Could any of that be relevant to your situation? Have you rebooted/reset all >> the network equipment between your server and the world? >> >> HTH >> >> ck >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see >> > -- ============================================ Brian Fristensky 971 Somerville Avenue Winnipeg MB R3T 1B4 CANADA bfristen at shaw.ca 204-261-3960 ============================================ From bfristen at shaw.ca Thu Sep 8 01:05:21 2011 From: bfristen at shaw.ca (Brian Fristensky) Date: Wed, 07 Sep 2011 20:05:21 -0500 Subject: [K12OSN] /etc/ltsp/dhcpd.conf must NOT be a symbolic link Message-ID: <4E6814D1.5020104@shaw.ca> I just installed the most recent LTSP on RHEL6.1, and everything went fine, except that ltsp-dhcpd would not start. After a lot of head scratching, I finally looked at /var/log/messages and discovered a "permission denied" message for dhcpd.conf. My practice is that I never change files in /etc, because I try to cut down on the size of things that need to be backed up. Rather, if I want to change a file, I put a copy of it in /usr/local/etc, which does get backed up, and then make a symbolic link to it. Rebuilding a system from scratch is as simple as recreating those symbolic links (a very small number) when a system is restored from backups. In any case, simply copying this file back to /etc/ltsp fixed it, and my thin client booted up and started ltsp flawlessly. It might be a useful tweak to ltsp to allow files to be symbolic links. Not sure why it didn't like the link. Brian From jomegat at jomegat.com Thu Sep 8 01:29:54 2011 From: jomegat at jomegat.com (Jomegat) Date: Wed, 07 Sep 2011 21:29:54 -0400 Subject: [K12OSN] /etc/ltsp/dhcpd.conf must NOT be a symbolic link In-Reply-To: <4E6814D1.5020104@shaw.ca> References: <4E6814D1.5020104@shaw.ca> Message-ID: <4E681A92.7070401@jomegat.com> On 09/07/2011 09:05 PM, Brian Fristensky wrote: > My practice is that I never change files in /etc, because I try to cut > down on > the size of things that need to be backed up. Rather, if I want to > change a file, I put a copy > of it in /usr/local/etc, which does get backed up, and then make a symbolic > link to it. Rebuilding a system from scratch is as simple as recreating > those > symbolic links (a very small number) when a system is restored from > backups. I've been toying with the idea of putting everything in /etc under svn control. Anyone have experience or thoughts on that? -- Jim Thomas jomegat at jomegat.com When you have little data, you have lots of freedom. - Olafur Ingolfssonr From microman at cmosnetworks.com Thu Sep 8 04:13:29 2011 From: microman at cmosnetworks.com (Terrell Prude' Jr.) Date: Thu, 08 Sep 2011 00:13:29 -0400 Subject: [K12OSN] /etc/ltsp/dhcpd.conf must NOT be a symbolic link In-Reply-To: <4E6814D1.5020104@shaw.ca> References: <4E6814D1.5020104@shaw.ca> Message-ID: <4E6840E9.7010105@cmosnetworks.com> Might be an SELinux thing, if you have it running in enforcing mode. I ran into this when moving virtual machines from one filesystem to another. Not only does the file need to be the right permissions, but so does the symlink ("chcon -h [security context] /my/symbolic/link" helps out here). --TP Brian Fristensky wrote: > I just installed the most recent LTSP on RHEL6.1, and everything went > fine, > except that ltsp-dhcpd would not start. > > After a lot of head scratching, I finally looked at /var/log/messages > and discovered a "permission denied" message for dhcpd.conf. > > My practice is that I never change files in /etc, because I try to cut > down on > the size of things that need to be backed up. Rather, if I want to > change a file, I put a copy > of it in /usr/local/etc, which does get backed up, and then make a > symbolic > link to it. Rebuilding a system from scratch is as simple as > recreating those > symbolic links (a very small number) when a system is restored from > backups. > > In any case, simply copying this file back to /etc/ltsp fixed it, and > my thin client > booted up and started ltsp flawlessly. > > It might be a useful tweak to ltsp to allow files to be symbolic > links. Not sure > why it didn't like the link. > > Brian > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see From news at siddall.name Thu Sep 8 13:58:19 2011 From: news at siddall.name (Jeff Siddall) Date: Thu, 08 Sep 2011 09:58:19 -0400 Subject: [K12OSN] depmod in chroot? Message-ID: <4E68C9FB.7020607@siddall.name> I have an x86_64 server with i686 clients. I run the chroot inside linux32 which does the trick for most things architecture related. However, I tried compiling and installing a chroot kernel module and had problems. Looks like autoconfig uses `uname -r` in a number of places and and of course that is the one place that still references x86_64. I was able to manually hack the Makefile and got the module compiled. However, the install process runs depmod -a and it is still looking for an x86_64 folder which of course does not exist. Any ideas how to get an i686 chroot kernel module installed on an x86_64 server? From jomegat at jomegat.com Thu Sep 8 14:25:07 2011 From: jomegat at jomegat.com (Jomegat) Date: Thu, 08 Sep 2011 10:25:07 -0400 Subject: [K12OSN] depmod in chroot? In-Reply-To: <4E68C9FB.7020607@siddall.name> References: <4E68C9FB.7020607@siddall.name> Message-ID: <4E68D043.8020401@jomegat.com> On 09/08/2011 09:58 AM, Jeff Siddall wrote: > I have an x86_64 server with i686 clients. I run the chroot inside > linux32 which does the trick for most things architecture related. > > However, I tried compiling and installing a chroot kernel module and had > problems. Looks like autoconfig uses `uname -r` in a number of places > and and of course that is the one place that still references x86_64. I > was able to manually hack the Makefile and got the module compiled. > However, the install process runs depmod -a and it is still looking for > an x86_64 folder which of course does not exist. > > Any ideas how to get an i686 chroot kernel module installed on an x86_64 > server? Do this as root: cd /opt/ltsp/i386 setarch i386 chroot . uname -a That should now say i686 rather than x86_64. Next you'll need to: mount /proc Then you can yum install whatever you want. Be sure to umount /proc before exiting the chroot shell: umount /proc exit If yum doesn't work, try copying /etc/resolv.conf to /opt/ltsp/i386/etc before the setarch step. This page: https://fedorahosted.org/k12linux/wiki/LocalAppsSupport shows an example of using setarch, but it leaves off the ever-important /etc/resolv.conf and more importantly, the mount /proc steps. -- Jim Thomas Principal Applications Engineer Bittware, Inc jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 In theory, theory and practice are the same, but in practice, they're not From news at siddall.name Thu Sep 8 14:49:57 2011 From: news at siddall.name (Jeff Siddall) Date: Thu, 08 Sep 2011 10:49:57 -0400 Subject: [K12OSN] depmod in chroot? In-Reply-To: <4E68D043.8020401@jomegat.com> References: <4E68C9FB.7020607@siddall.name> <4E68D043.8020401@jomegat.com> Message-ID: <4E68D615.8070405@siddall.name> On 09/08/2011 10:25 AM, Jomegat wrote: > On 09/08/2011 09:58 AM, Jeff Siddall wrote: >> I have an x86_64 server with i686 clients. I run the chroot inside >> linux32 which does the trick for most things architecture related. >> >> However, I tried compiling and installing a chroot kernel module and had >> problems. Looks like autoconfig uses `uname -r` in a number of places >> and and of course that is the one place that still references x86_64. I >> was able to manually hack the Makefile and got the module compiled. >> However, the install process runs depmod -a and it is still looking for >> an x86_64 folder which of course does not exist. >> >> Any ideas how to get an i686 chroot kernel module installed on an x86_64 >> server? > > Do this as root: > cd /opt/ltsp/i386 > setarch i386 chroot . > uname -a > > That should now say i686 rather than x86_64. Not really: # cd /opt/ltsp/i386 # setarch i386 chroot . bash-4.1# uname -a Linux hostname 2.6.32-131.6.1.el6.x86_64 #1 SMP Tue Jul 12 17:14:50 CDT 2011 i686 i686 i386 GNU/Linux Yes, the machine hardware shows i686: bash-4.1# uname -m i686 But the kernel still shows x86_64: bash-4.1# uname -r 2.6.32-131.6.1.el6.x86_64 And depmod is, predictably, looking at the running kernel: bash-4.1# depmod -a WARNING: Couldn't open directory /lib/modules/2.6.32-131.6.1.el6.x86_64: No such file or directory FATAL: Could not open /lib/modules/2.6.32-131.6.1.el6.x86_64/modules.dep.temp for writing: No such file or directory However, after some more reading I discovered it is possible to override the running kernel by adding a specific version as a parameter. Not sure how I missed that when googling. I guess I was searching for "arch" instead of "version". Anyway, problem solved for now. Still, if anyone knows a way to override what uname -r returns so that cross-compiling "just works" in the chroot that would be a lot nicer. From reb at taco.com Fri Sep 9 14:45:22 2011 From: reb at taco.com (Phydeaux) Date: Fri, 9 Sep 2011 10:45:22 -0400 Subject: [K12OSN] Firefox core dumps - CentOS 6 Message-ID: <95ffc5d88295ac18f4d0d68700e29379.squirrel@webmail.taco.com> We've moved our LTSP installation from Fedora 13 to CentOS 6 over the summer and school is set to start Monday. All seemed well, but then we fired up Firefox. After starting and displaying a window (sometimes it even renders the window) it core dumps with a bus error as follows: /usr/lib64/firefox-3.6/run-mozilla.sh: line 131: 32095 Bus error (core dumped) "$prog" ${1+"$@"} We tried installing Google Chrome, but it too fails to load, displaying this: /opt/google/chrome/chrome: /lib64/libz.so.1: no version information available (required by /opt/google/chrome/chrome) [31429:31429:606701736390:ERROR:browser_main.cc(970)] GLib-GObject: gsignal.c:2270: signal `preedit-changed' is invalid for instance `0x578fe30' Bus error (core dumped) We've got a plain-vanilla CentOS installation. I tried removing the .mozilla directory to allow it to start from scratch. I tried both with and without the Flash plugin installed. I've gotten nowhere. Based on research and reports of similar problems I found I have reinstalled xulrunner, JRE, Firefox, Chrome, and probably a few other packages School starts Monday. Does anyone have any clues as to what might be wrong? reb From dean.jones at oregonstate.edu Fri Sep 9 15:19:25 2011 From: dean.jones at oregonstate.edu (Dean Jones) Date: Fri, 9 Sep 2011 08:19:25 -0700 Subject: [K12OSN] Firefox core dumps - CentOS 6 In-Reply-To: <95ffc5d88295ac18f4d0d68700e29379.squirrel@webmail.taco.com> References: <95ffc5d88295ac18f4d0d68700e29379.squirrel@webmail.taco.com> Message-ID: Are the clients thin or fat? I have had no issues with FF on CentOS 6, just FYI. On Fri, Sep 9, 2011 at 7:45 AM, Phydeaux wrote: > We've moved our LTSP installation from Fedora 13 to CentOS 6 over the summer > and school is set to start Monday. ?All seemed well, but then we fired up > Firefox. ?After starting and displaying a window (sometimes it even renders > the window) it core dumps with a bus error as follows: > > ? /usr/lib64/firefox-3.6/run-mozilla.sh: line 131: 32095 Bus error (core dumped) > "$prog" ${1+"$@"} > > We tried installing Google Chrome, but it too fails to load, displaying this: > > ? /opt/google/chrome/chrome: /lib64/libz.so.1: no version information available > (required by /opt/google/chrome/chrome) > ? [31429:31429:606701736390:ERROR:browser_main.cc(970)] GLib-GObject: > gsignal.c:2270: signal `preedit-changed' is invalid for instance `0x578fe30' > ? Bus error (core dumped) > > We've got a plain-vanilla CentOS installation. ?I tried removing the .mozilla > directory to allow it to start from scratch. ?I tried both with and without > the Flash plugin installed. ?I've gotten nowhere. ?Based on research and > reports of similar problems I found I have reinstalled xulrunner, JRE, Firefox, > Chrome, and probably a few other packages > > School starts Monday. ?Does anyone have any clues as to what might be wrong? > > reb > > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > From reb at taco.com Fri Sep 9 15:48:36 2011 From: reb at taco.com (Phydeaux) Date: Fri, 9 Sep 2011 11:48:36 -0400 Subject: [K12OSN] Firefox core dumps - CentOS 6 In-Reply-To: References: <95ffc5d88295ac18f4d0d68700e29379.squirrel@webmail.taco.com> Message-ID: <879e1851756e67ca80cefbec80157427.squirrel@webmail.taco.com> > Are the clients thin or fat? The clients are thin (but pretty beefy) but it doesn't even take a client to do this. I ssh'd in and sent the X session back to the box on my desk and see the same thing. We've got a plain-vanilla CentOS 6 install. reb From dean.jones at oregonstate.edu Fri Sep 9 16:24:37 2011 From: dean.jones at oregonstate.edu (Dean Jones) Date: Fri, 9 Sep 2011 09:24:37 -0700 Subject: [K12OSN] Firefox core dumps - CentOS 6 In-Reply-To: <879e1851756e67ca80cefbec80157427.squirrel@webmail.taco.com> References: <95ffc5d88295ac18f4d0d68700e29379.squirrel@webmail.taco.com> <879e1851756e67ca80cefbec80157427.squirrel@webmail.taco.com> Message-ID: Maybe this will help This is what I have on my CentOS 6 boxes from the repos. firefox.x86_64 3.6.9-2.el6.centos @base and the version of the dependencies for firefox: alsa-lib.x86_64 1.0.21-3.el6 @base atk.x86_64 1.28.0-2.el6 @base bash.x86_64 4.1.2-3.el6 @anaconda-CentOS-201106060106.x86_64/6.0 cairo.x86_64 1.8.8-3.1.el6 @base centos-indexhtml.noarch 6-1.el6.centos @base fontconfig.x86_64 2.8.0-3.el6 @base freetype.x86_64 2.3.11-6.el6_0.2 @updates glib2.x86_64 2.22.5-5.el6 @anaconda-CentOS-201106060106.x86_64/6.0 glibc.x86_64 2.12-1.7.el6_0.5 @updates gtk2.x86_64 2.18.9-4.el6 @base libcanberra-gtk2.x86_64 0.22-1.el6 @base libgcc.x86_64 4.4.4-13.el6 @anaconda-CentOS-201106060106.x86_64/6.0 libstdc++.x86_64 4.4.4-13.el6 @anaconda-CentOS-201106060106.x86_64/6.0 nspr.x86_64 4.8.6-1.el6 @anaconda-CentOS-201106060106.x86_64/6.0 pango.x86_64 1.28.1-3.el6_0.5 @updates pulseaudio-libs-glib2.x86_64 0.9.21-13.el6 @base pycairo.x86_64 1.8.6-2.1.el6 @base pygtk2.x86_64 2.16.0-3.el6 @base redhat-bookmarks.noarch 6-1.el6.centos @base startup-notification.x86_64 0.10-2.1.el6 @base xulrunner.x86_64 1.9.2.9-1.el6.centos @base On Fri, Sep 9, 2011 at 8:48 AM, Phydeaux wrote: >> Are the clients thin or fat? > > The clients are thin (but pretty beefy) but it doesn't even take a client > to do this. ?I ssh'd in and sent the X session back to the box on my desk > and see the same thing. ?We've got a plain-vanilla CentOS 6 install. > > reb > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > From reb at taco.com Fri Sep 9 17:20:54 2011 From: reb at taco.com (Phydeaux) Date: Fri, 9 Sep 2011 13:20:54 -0400 Subject: [K12OSN] Firefox core dumps - CentOS 6 In-Reply-To: References: <95ffc5d88295ac18f4d0d68700e29379.squirrel@webmail.taco.com> <879e1851756e67ca80cefbec80157427.squirrel@webmail.taco.com> Message-ID: <21a5c8ea5574a509bd1dc51ce43d216a.squirrel@webmail.taco.com> > Maybe this will help > > This is what I have on my CentOS 6 boxes from the repos. Thanks -- I had all of those installed, and at the same versions and from the same repositories. I flushed the info yum was storing with 'yum clean all' and reinstalled all of them. There was no change in behaviour. I do note that my firefox dependency tree did not have these packages listed: libcanberra-gtk2 pulseaudio-libs-glib2 pycairo pygtk2 After some more poking around I realized that I hadn't properly disabled the Adobe flash plugin. After removing the link in /usr/lib64/mozilla/plugins to /opt/adobe/flash/libflashplayer11_b2_install_lin_64_080811.so firefox worked once again. Which flash plugin should I be using? Clearly this one isn't working. We've got all 64-bit hardware, but are running 32-bit LTSP thin clients without local apps. reb From dean.jones at oregonstate.edu Fri Sep 9 18:01:49 2011 From: dean.jones at oregonstate.edu (Dean Jones) Date: Fri, 9 Sep 2011 11:01:49 -0700 Subject: [K12OSN] Firefox core dumps - CentOS 6 In-Reply-To: <21a5c8ea5574a509bd1dc51ce43d216a.squirrel@webmail.taco.com> References: <95ffc5d88295ac18f4d0d68700e29379.squirrel@webmail.taco.com> <879e1851756e67ca80cefbec80157427.squirrel@webmail.taco.com> <21a5c8ea5574a509bd1dc51ce43d216a.squirrel@webmail.taco.com> Message-ID: Hi, I am using the 64 bit Flash plugin. It is downloadable from Adobe's site and is in beta or some such. But it works and does not break FF. My think clients are the same as yours, 32 bit, no local apps. So FF is running on the x86_64 server so the 64bit flash would be best. On Fri, Sep 9, 2011 at 10:20 AM, Phydeaux wrote: >> Maybe this will help >> >> This is what I have on my CentOS 6 boxes from the repos. > > Thanks -- I had all of those installed, and at the same versions and from > the same repositories. ?I flushed the info yum was storing with 'yum clean all' > and reinstalled all of them. ?There was no change in behaviour. ?I do note that > my firefox dependency tree did not have these packages listed: > > libcanberra-gtk2 > pulseaudio-libs-glib2 > pycairo > pygtk2 > > After some more poking around I realized that I hadn't properly disabled the > Adobe flash plugin. ?After removing the link in /usr/lib64/mozilla/plugins to > /opt/adobe/flash/libflashplayer11_b2_install_lin_64_080811.so firefox worked > once again. > > Which flash plugin should I be using? ?Clearly this one isn't working. We've > got all 64-bit hardware, but are running 32-bit LTSP thin clients without > local apps. > > reb > > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > From reb at taco.com Fri Sep 9 18:48:10 2011 From: reb at taco.com (Phydeaux) Date: Fri, 9 Sep 2011 14:48:10 -0400 Subject: [K12OSN] Firefox core dumps - CentOS 6 In-Reply-To: References: <95ffc5d88295ac18f4d0d68700e29379.squirrel@webmail.taco.com> <879e1851756e67ca80cefbec80157427.squirrel@webmail.taco.com> <21a5c8ea5574a509bd1dc51ce43d216a.squirrel@webmail.taco.com> Message-ID: <55343381fdc746d1e57bff117c08311c.squirrel@webmail.taco.com> > I am using the 64 bit Flash plugin. It is downloadable from Adobe's > site and is in beta or some such. But it works and does not break FF. > > My think clients are the same as yours, 32 bit, no local apps. So FF > is running on the x86_64 server so the 64bit flash would be best. > That's what I figured. I have no idea which version was installed before, or where it came from. I grabbed the latest one from Adobe and now all seems well. My browser now reports this on the Adobe Flash test page: You have version 11,0,1,129 installed Thanks! reb From ahodson at elp.rr.com Sat Sep 10 00:07:30 2011 From: ahodson at elp.rr.com (Alan Hodson) Date: Fri, 09 Sep 2011 18:07:30 -0600 Subject: [K12OSN] OT: Anybody working with edubuntu 11:04 and stateless TC's? Message-ID: <4E6AAA42.3070500@elp.rr.com> Hi folks I am working with a school that has a Dell PowerEdge T105 server, and the classic LTSP configuration for thin clients keeps eluding me. I must be skipping a step or overlooking an install for the thin clients to light up. Any personal anecdotes someone might want to share? I'm looking for the "make sure you ..." type of comment. Thanks A.Hodson HodsonDTS.org From dahopkins429 at gmail.com Sun Sep 11 03:31:53 2011 From: dahopkins429 at gmail.com (David Hopkins) Date: Sat, 10 Sep 2011 23:31:53 -0400 Subject: [K12OSN] OT: Anybody working with edubuntu 11:04 and stateless TC's? In-Reply-To: <4E6AAA42.3070500@elp.rr.com> References: <4E6AAA42.3070500@elp.rr.com> Message-ID: Not sure how far you have gotten or where the process fails for you so ... I've use Lucid 10.04 at the moment but have tried the 11.04 just because of the LTSP 5.2 support. (look forward to the Centos6 version!) There didn't seem to be anything special about the install. Just be certain that you have the correct ltsp packages installed? I check by simply using synaptic and filtering on ltsp to see what packages are installed. Next, I'd make sure that the network is correct, ifconfig should show both interfaces configured. The only thing I definitely do is to remove the Network Manager and manually define the settings in the /etc/network/interfaces configuration file. If ltsp is installed, check that you have the correct client image. My systems tend to be AMD based, so I have to explicitly build the client i386 images with ltsp-build-client --arch i386. If it completes correctly, you should have a /var/lib/tftpboot/ltsp/i386 directory and also /opt/ltsp/images with the image. This is enough to just boot a client. Also, check that the clients are set to PXE boot. Sincerely, Dave Hopkins On Fri, Sep 9, 2011 at 8:07 PM, Alan Hodson wrote: > Hi folks > > I am working with a school that has a Dell PowerEdge T105 server, and the > classic LTSP configuration for thin clients keeps eluding me. I must be > skipping a step or overlooking an install for the thin clients to light up. > Any personal anecdotes someone might want to share? I'm looking for the > "make sure you ..." type of comment. > Thanks > A.Hodson > HodsonDTS.org > > ______________________________**_________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/**mailman/listinfo/k12osn > For more info see > -------------- next part -------------- An HTML attachment was scrubbed... URL: From jan at recreatie-zorg.nl Mon Sep 12 07:55:53 2011 From: jan at recreatie-zorg.nl (Jan Middelkoop) Date: Mon, 12 Sep 2011 09:55:53 +0200 Subject: [K12OSN] OT: Anybody working with edubuntu 11:04 and stateless TC's? In-Reply-To: <4E6AAA42.3070500@elp.rr.com> References: <4E6AAA42.3070500@elp.rr.com> Message-ID: <4E6DBB09.1090606@recreatie-zorg.nl> Hi, For Ubuntu support, you should really visit #ltsp on irc.freenode.net. There are lots of Ubuntu LTSP users there. Kindest regards, Jan Middelkoop Op 10-9-2011 2:07, Alan Hodson schreef: > Hi folks > > I am working with a school that has a Dell PowerEdge T105 server, and > the classic LTSP configuration for thin clients keeps eluding me. I > must be skipping a step or overlooking an install for the thin clients > to light up. Any personal anecdotes someone might want to share? I'm > looking for the "make sure you ..." type of comment. > Thanks > A.Hodson > HodsonDTS.org > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see From ltsp at symbio-technologies.com Mon Sep 12 15:32:52 2011 From: ltsp at symbio-technologies.com (Gideon Romm) Date: Mon, 12 Sep 2011 11:32:52 -0400 Subject: [K12OSN] Resend: Thin client parts on offer Message-ID: Hello, all. I know for many of you, September marks the end of summer and beginning of a new year, and though Spring is usually for cleaning, we have chosen to make some room now. So, I have a partial list of odds and ends that may be of interest to someone on this list. I know many folks here may look for a way to get a few more thin clients at a reduced price. Consider the prices on this list a suggested price. We would like to get all this stuff cleared out by the end of the month, so make us an offer. Prices obviously don't include shipping and I am hoping they will find a home in North America (we are in New York), so as not to have to worry about the various duties associated with selling abroad. Here is a link to the list: http://symbio-technologies.com/LTSP-Sale-201109-total.pdf As always, you all know where to find me. Happy New Year! :) -Gadi -- Gideon Romm Symbio Technologies | www.symbio-technologies.com (o) 914.576.1205 | (f) 914.576.0944 From jlunning at new.org Mon Sep 12 16:40:21 2011 From: jlunning at new.org (Justin Lunning) Date: Mon, 12 Sep 2011 12:40:21 -0400 Subject: [K12OSN] drm i915_hangcheck_elapsed on client In-Reply-To: <4E678D51.4020807@astrofoto.org> References: <4E678D51.4020807@astrofoto.org> Message-ID: <4E6E35F5.4060008@new.org> Hi Roland, I've seen a similar message too many times to count. I don't use RHEL, but have seen it a lot with Ubuntu/Debian LTSP setups that we deploy. There are bugs in the Linux Intel Graphics driver that cause this. There are a number of adjustments you can make to try and remedy this, including upgrading/downgrading both the kernel and the intel graphics driver and adjusting acceleration options in custom xorg.conf files for your thin client, but it's a mixed bag. You basically just have to try different combinations of adjustments until you find something that works. Different techniques seem to have different mileage on different hardware. It can be very time consuming. If you google the error message you'll find a lot of info about. Justin Lunning npServ Director NEW (Nonprofit Enterprise at Work) jlunning at new.org / www.new.org (734) 998-0160 x 212 (Ann Arbor) (313) 887-7788 x 212 (Detroit) Ann Arbor: The NEW Center, 1100 N. Main Street Detroit: Hannan House, 4750 Woodward Ave HELPING NONPROFITS SUCCEED On 9/7/2011 11:27 AM, Roland Roberts wrote: > I've had my client freeze up, temporarily, a number of times. Not > every day, not with high-frequency, but more than once a week. Today, > after it recovered, I launched a local xterm and found this in > /var/log/messages > > Sep 7 11:05:08 client-192 kernel: [drm:i915_hangcheck_elapsed] > *ERROR* Hangcheck timer elapsed... GPU idle, missed IRQ. > > The previous log message is from the previous evening, the next if > from me lauching the local xterm. > > The client is a Thinkpad 420s with a Core i5 and the integrated Intel > GPU. Since that's a new addition for RHEL/SL 6.1, it's probably not > client related, but I thought I'd bounce it around here first. > > roland > From toddobryan at gmail.com Mon Sep 19 02:21:23 2011 From: toddobryan at gmail.com (Todd O'Bryan) Date: Sun, 18 Sep 2011 22:21:23 -0400 Subject: [K12OSN] checking out equipment Message-ID: Hey all, The journalism department at my school needs a way to keep track of digital cameras, microphones, video cameras, etc., and asked if I had any ideas. A library circulation program is probably more than they need, but they would like to be able to barcode equipment and scan student ids when they check stuff out. Does anyone have a suggestion for a program that might be appropriate? Todd From burke at thealmquists.net Mon Sep 19 04:22:00 2011 From: burke at thealmquists.net (Burke Almquist) Date: Sun, 18 Sep 2011 23:22:00 -0500 Subject: [K12OSN] checking out equipment In-Reply-To: References: Message-ID: <557FC8E6-ABEE-4D16-A638-4AA4B2F28186@thealmquists.net> Does your school have an existing library system that they could simply piggy-back onto? On Sep 18, 2011, at 9:21 PM, Todd O'Bryan wrote: > Hey all, > > The journalism department at my school needs a way to keep track of > digital cameras, microphones, video cameras, etc., and asked if I had > any ideas. A library circulation program is probably more than they > need, but they would like to be able to barcode equipment and scan > student ids when they check stuff out. > > Does anyone have a suggestion for a program that might be appropriate? > > Todd > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see From toddobryan at gmail.com Tue Sep 20 02:35:49 2011 From: toddobryan at gmail.com (Todd O'Bryan) Date: Mon, 19 Sep 2011 22:35:49 -0400 Subject: [K12OSN] checking out equipment In-Reply-To: <557FC8E6-ABEE-4D16-A638-4AA4B2F28186@thealmquists.net> References: <557FC8E6-ABEE-4D16-A638-4AA4B2F28186@thealmquists.net> Message-ID: The equipment doesn't live in the library, so that would be a little complicated. I found GCStar and they're going to try that. Since it's only 100 or so kids who'd do the checking out, that may be all they need. On Mon, Sep 19, 2011 at 12:22 AM, Burke Almquist wrote: > Does your school have an existing library system that they could simply piggy-back onto? > > On Sep 18, 2011, at 9:21 PM, Todd O'Bryan wrote: > >> Hey all, >> >> The journalism department at my school needs a way to keep track of >> digital cameras, microphones, video cameras, etc., and asked if I had >> any ideas. A library circulation program is probably more than they >> need, but they would like to be able to barcode equipment and scan >> student ids when they check stuff out. >> >> Does anyone have a suggestion for a program that might be appropriate? >> >> Todd >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see > > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > From jomegat at jomegat.com Tue Sep 20 02:49:39 2011 From: jomegat at jomegat.com (Jomegat) Date: Mon, 19 Sep 2011 22:49:39 -0400 Subject: [K12OSN] checking out equipment In-Reply-To: References: <557FC8E6-ABEE-4D16-A638-4AA4B2F28186@thealmquists.net> Message-ID: <4E77FF43.5000704@jomegat.com> You could also look at OpenBiblio if that doesn't work out. It's a library OPAC, but it is far simpler than something like Koha. On 09/19/2011 10:35 PM, Todd O'Bryan wrote: > The equipment doesn't live in the library, so that would be a little > complicated. > > I found GCStar and they're going to try that. Since it's only 100 or > so kids who'd do the checking out, that may be all they need. > > On Mon, Sep 19, 2011 at 12:22 AM, Burke Almquist wrote: >> Does your school have an existing library system that they could simply piggy-back onto? >> >> On Sep 18, 2011, at 9:21 PM, Todd O'Bryan wrote: >> >>> Hey all, >>> >>> The journalism department at my school needs a way to keep track of >>> digital cameras, microphones, video cameras, etc., and asked if I had >>> any ideas. A library circulation program is probably more than they >>> need, but they would like to be able to barcode equipment and scan >>> student ids when they check stuff out. >>> >>> Does anyone have a suggestion for a program that might be appropriate? >>> >>> Todd >>> >>> _______________________________________________ >>> K12OSN mailing list >>> K12OSN at redhat.com >>> https://www.redhat.com/mailman/listinfo/k12osn >>> For more info see >> >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see >> > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > -- Jim Thomas jomegat at jomegat.com Build a man a fire and he is warm for a day. Set a man on fire and he is warm for the rest of his life. From jomegat at jomegat.com Tue Sep 20 19:18:08 2011 From: jomegat at jomegat.com (Jomegat) Date: Tue, 20 Sep 2011 15:18:08 -0400 Subject: [K12OSN] Poor performance woes Message-ID: <4E78E6F0.9090207@jomegat.com> I have a K12Linux setup based on EL6-64. The TC's are ancient i686's with 256M RAM and 10/100 nics. The TS has a gigE NIC, and between the TS & TCs is a managed switch with 2 gigE ports and 24 10/100 ports. The TS is connected via one of the gigE's. We have a student who is having real performance problems when he loads a page. I went in today to be there while he was experiencing woe, and ran the system monitor applet. When he loads the web page - with no flash in it that I can find - the network maxes out at 12MiB/sec, and it takes an eternity for the page to load. He reports that sometimes this particular page will cause the TC to crash (black screen with text followed by login screen), but he was not able to reproduce that while I was there. I suspect there is some user error involved in the crash scenario, but it will be difficult to prove. I had tried running firefox as a local app last month, but the school needs flash, and it requires 512M. Performance on these TC's was substantially worse when running FF as a local app, so I backed off of that. The page he was trying to load is password protected, but I had him ctrl-U while he had it open, and save the source to a file which I can share if need be. It is filled with javascript, but I'm not sure that's the culprit (though it seems the most likely suspect). I found no .flv, FLV, .swf, or .SWF in the file anywhere. As an experiment, I connected a monitor, keybd, and mouse directly to the server and had him login there. The page loads just fine that way. That might be a possible solution, but for some reason, the display is incredibly blurry. If I make that permanent, I'd have to run some cable through the wall (they're pinched in the door right now, and barely reach). Before they called me in to look at this, he had been bringing his mom's laptop to school so he could do his work (it's an online course). If it crashes while he is taking a quiz, he loses a test attempt or gets a bad grade, so this has to be reliable. His mom is on the school board, and she is pushing the school to abandon LTSP in favor of a Windows solution. They think that if they install some flavor of windows on the TC's they will get better performance, but I believe they won't, as flash will need 512M no matter the OS. I think they will need all new PC's if they go that route. I am on the cusp of losing my network to Windows, at which point I will resign my post as unpaid volunteer sysadmin, so if I sound a little desperate, now you know why. Any help here would be greatly appreciated. -- Jim Thomas Principal Applications Engineer Bittware, Inc jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 The problem with the future is that it keeps turning into the present - Hobbes From dahopkins429 at gmail.com Tue Sep 20 20:12:34 2011 From: dahopkins429 at gmail.com (David Hopkins) Date: Tue, 20 Sep 2011 16:12:34 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E78E6F0.9090207@jomegat.com> References: <4E78E6F0.9090207@jomegat.com> Message-ID: So, runs fine at server console, but lags severely at the thin client with the same account. Are you certain that the clients are running at 100Mb full duplex? If you've enabled a shell at the client can you check to see what ifconfig reports at the client? I would check memory/performance at the client with top (in a shell, add SCREEN_02=shell in lts.conf, ctl-alt-f2 to access it at the client) while the page is being displayed from the GUI login session just to convince myself it isn't a load issue at the client. Also, Adobe has a 64bit Flash for linux (v11) available as beta. I've been running that for the last couple of weeks (since it was released IOW) on my systems (Ubuntu 10.04 with LTSP packages added). http://labs.adobe.com/downloads/flashplayer11.html It definitely runs better than the 32bit flash with nswrapper. Installation was just a matter of uninstalling the 32bit flash and copying the 64bit version to the correct plugins directory. Those are the things that I'd check first. Sincerely, Dave Hopkins Newark Charter School On Tue, Sep 20, 2011 at 3:18 PM, Jomegat wrote: > I have a K12Linux setup based on EL6-64. The TC's are ancient i686's with > 256M RAM and 10/100 nics. The TS has a gigE NIC, and between the TS & TCs > is a managed switch with 2 gigE ports and 24 10/100 ports. The TS is > connected via one of the gigE's. > > We have a student who is having real performance problems when he loads a > page. I went in today to be there while he was experiencing woe, and ran > the system monitor applet. > > When he loads the web page - with no flash in it that I can find - the > network maxes out at 12MiB/sec, and it takes an eternity for the page to > load. He reports that sometimes this particular page will cause the TC to > crash (black screen with text followed by login screen), but he was not able > to reproduce that while I was there. I suspect there is some user error > involved in the crash scenario, but it will be difficult to prove. > > I had tried running firefox as a local app last month, but the school needs > flash, and it requires 512M. Performance on these TC's was substantially > worse when running FF as a local app, so I backed off of that. > > The page he was trying to load is password protected, but I had him ctrl-U > while he had it open, and save the source to a file which I can share if > need be. It is filled with javascript, but I'm not sure that's the culprit > (though it seems the most likely suspect). I found no .flv, FLV, .swf, or > .SWF in the file anywhere. > > As an experiment, I connected a monitor, keybd, and mouse directly to the > server and had him login there. The page loads just fine that way. That > might be a possible solution, but for some reason, the display is incredibly > blurry. If I make that permanent, I'd have to run some cable through the > wall (they're pinched in the door right now, and barely reach). > > Before they called me in to look at this, he had been bringing his mom's > laptop to school so he could do his work (it's an online course). If it > crashes while he is taking a quiz, he loses a test attempt or gets a bad > grade, so this has to be reliable. > > His mom is on the school board, and she is pushing the school to abandon > LTSP in favor of a Windows solution. They think that if they install some > flavor of windows on the TC's they will get better performance, but I > believe they won't, as flash will need 512M no matter the OS. I think they > will need all new PC's if they go that route. > > I am on the cusp of losing my network to Windows, at which point I will > resign my post as unpaid volunteer sysadmin, so if I sound a little > desperate, now you know why. > > Any help here would be greatly appreciated. > > -- > Jim Thomas Principal Applications Engineer Bittware, Inc > jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 > The problem with the future is that it keeps turning into the present - > Hobbes > > ______________________________**_________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/**mailman/listinfo/k12osn > For more info see > -------------- next part -------------- An HTML attachment was scrubbed... URL: From jomegat at jomegat.com Tue Sep 20 20:23:02 2011 From: jomegat at jomegat.com (Jomegat) Date: Tue, 20 Sep 2011 16:23:02 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: References: <4E78E6F0.9090207@jomegat.com> Message-ID: <4E78F626.3090104@jomegat.com> On 09/20/2011 04:12 PM, David Hopkins wrote: > So, runs fine at server console, but lags severely at the thin client > with the same account. Are you certain that the clients are running at > 100Mb full duplex? If you've enabled a shell at the client can you > check to see what ifconfig reports at the client? I would check > memory/performance at the client with top (in a shell, add > SCREEN_02=shell in lts.conf, ctl-alt-f2 to access it at the client) > while the page is being displayed from the GUI login session just to > convince myself it isn't a load issue at the client. I will double check, but since I was seeing 12MiB/sec - which is pretty close to 100 Mbits/sec, I'm thinking they are running at 100. Also, the LEDs on the switch say 100, and not 10. > > Also, Adobe has a 64bit Flash for linux (v11) available as beta. I've > been running that for the last couple of weeks (since it was released > IOW) on my systems (Ubuntu 10.04 with LTSP packages added). > http://labs.adobe.com/downloads/flashplayer11.html It definitely runs > better than the 32bit flash with nswrapper. Installation was just a > matter of uninstalling the 32bit flash and copying the 64bit version to > the correct plugins directory. Yes, I'm using the 64-bit Flash, and not the nswrapper version. Further, I can't find any flash content in the page he's loading, just a truckload of javascript. > > Those are the things that I'd check first. Thanks. > > Sincerely, > Dave Hopkins > Newark Charter School > > On Tue, Sep 20, 2011 at 3:18 PM, Jomegat > wrote: > > I have a K12Linux setup based on EL6-64. The TC's are ancient > i686's with 256M RAM and 10/100 nics. The TS has a gigE NIC, and > between the TS & TCs is a managed switch with 2 gigE ports and 24 > 10/100 ports. The TS is connected via one of the gigE's. > > We have a student who is having real performance problems when he > loads a page. I went in today to be there while he was experiencing > woe, and ran the system monitor applet. > > When he loads the web page - with no flash in it that I can find - > the network maxes out at 12MiB/sec, and it takes an eternity for the > page to load. He reports that sometimes this particular page will > cause the TC to crash (black screen with text followed by login > screen), but he was not able to reproduce that while I was there. I > suspect there is some user error involved in the crash scenario, but > it will be difficult to prove. > > I had tried running firefox as a local app last month, but the > school needs flash, and it requires 512M. Performance on these TC's > was substantially worse when running FF as a local app, so I backed > off of that. > > The page he was trying to load is password protected, but I had him > ctrl-U while he had it open, and save the source to a file which I > can share if need be. It is filled with javascript, but I'm not > sure that's the culprit (though it seems the most likely suspect). > I found no .flv, FLV, .swf, or .SWF in the file anywhere. > > As an experiment, I connected a monitor, keybd, and mouse directly > to the server and had him login there. The page loads just fine > that way. That might be a possible solution, but for some reason, > the display is incredibly blurry. If I make that permanent, I'd > have to run some cable through the wall (they're pinched in the door > right now, and barely reach). > > Before they called me in to look at this, he had been bringing his > mom's laptop to school so he could do his work (it's an online > course). If it crashes while he is taking a quiz, he loses a test > attempt or gets a bad grade, so this has to be reliable. > > His mom is on the school board, and she is pushing the school to > abandon LTSP in favor of a Windows solution. They think that if > they install some flavor of windows on the TC's they will get better > performance, but I believe they won't, as flash will need 512M no > matter the OS. I think they will need all new PC's if they go that > route. > > I am on the cusp of losing my network to Windows, at which point I > will resign my post as unpaid volunteer sysadmin, so if I sound a > little desperate, now you know why. > > Any help here would be greatly appreciated. > > -- > Jim Thomas Principal Applications Engineer Bittware, Inc > jthomas at bittware.com > http://www.bittware.com (603) 226-0404 x536 > > The problem with the future is that it keeps turning into the > present - Hobbes > > _________________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/__mailman/listinfo/k12osn > > For more info see > > > > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see -- Jim Thomas Principal Applications Engineer Bittware, Inc jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 The problem with the future is that it keeps turning into the present - Hobbes From dean.jones at oregonstate.edu Tue Sep 20 20:34:16 2011 From: dean.jones at oregonstate.edu (Dean Jones) Date: Tue, 20 Sep 2011 13:34:16 -0700 Subject: [K12OSN] Poor performance woes In-Reply-To: References: <4E78E6F0.9090207@jomegat.com> Message-ID: +1 on the 64 bit flash. Much better performance if the problem is flash related. That said, i would guess that the main issue is that your 100meg connection is maxing out and going as fast as it can. That 12MiB/sec is as good as you are going to get. X is very bandwidth heavy and the network is maxed. I would guess that the webpage is doing something rather intense (and dumb?) with redrawing or something display related and is just eating everything up. I have similarly speced terminal clients as you describe and I can choke them with some certain display heavy applications but going to a gig connection makes them (more) usable. If the clients support that it would be an inexpensive upgrade option. As you know Windows will perform poorly on those clients as well regardless of the server specs. It is dated hardware with little video memory. You can try nbdroot for a slight speed increase over NFS if the problem is file i/o related. On Tue, Sep 20, 2011 at 1:12 PM, David Hopkins wrote: > So, runs fine at server console, but lags severely at the thin client with > the same account.? Are you certain that the clients are running at 100Mb > full duplex?? If you've enabled a shell at the client can you check to see > what ifconfig reports at the client?? I would check memory/performance at > the client with top (in a shell, add SCREEN_02=shell in lts.conf, ctl-alt-f2 > to access it at the client) while the page is being displayed from the GUI > login session just to convince myself it isn't a load issue at the client. > > Also, Adobe has a 64bit Flash for linux (v11) available as beta. I've been > running that for the last couple of weeks (since it was released IOW) on my > systems (Ubuntu 10.04 with LTSP packages added). > http://labs.adobe.com/downloads/flashplayer11.html It definitely runs better > than the 32bit flash with nswrapper.? Installation was just a matter of > uninstalling the 32bit flash and copying the 64bit version to the correct > plugins directory. > > Those are the things that I'd check first. > > Sincerely, > Dave Hopkins > Newark Charter School > > On Tue, Sep 20, 2011 at 3:18 PM, Jomegat wrote: >> >> I have a K12Linux setup based on EL6-64. ?The TC's are ancient i686's with >> 256M RAM and 10/100 nics. ?The TS has a gigE NIC, and between the TS & TCs >> is a managed switch with 2 gigE ports and 24 10/100 ports. ?The TS is >> connected via one of the gigE's. >> >> We have a student who is having real performance problems when he loads a >> page. ?I went in today to be there while he was experiencing woe, and ran >> the system monitor applet. >> >> When he loads the web page - with no flash in it that I can find - the >> network maxes out at 12MiB/sec, and it takes an eternity for the page to >> load. ?He reports that sometimes this particular page will cause the TC to >> crash (black screen with text followed by login screen), but he was not able >> to reproduce that while I was there. ?I suspect there is some user error >> involved in the crash scenario, but it will be difficult to prove. >> >> I had tried running firefox as a local app last month, but the school >> needs flash, and it requires 512M. ?Performance on these TC's was >> substantially worse when running FF as a local app, so I backed off of that. >> >> The page he was trying to load is password protected, but I had him ctrl-U >> while he had it open, and save the source to a file which I can share if >> need be. ?It is filled with javascript, but I'm not sure that's the culprit >> (though it seems the most likely suspect). ?I found no .flv, FLV, .swf, or >> .SWF in the file anywhere. >> >> As an experiment, I connected a monitor, keybd, and mouse directly to the >> server and had him login there. ?The page loads just fine that way. ?That >> might be a possible solution, but for some reason, the display is incredibly >> blurry. ?If I make that permanent, I'd have to run some cable through the >> wall (they're pinched in the door right now, and barely reach). >> >> Before they called me in to look at this, he had been bringing his mom's >> laptop to school so he could do his work (it's an online course). ?If it >> crashes while he is taking a quiz, he loses a test attempt or gets a bad >> grade, so this has to be reliable. >> >> His mom is on the school board, and she is pushing the school to abandon >> LTSP in favor of a Windows solution. ?They think that if they install some >> flavor of windows on the TC's they will get better performance, but I >> believe they won't, as flash will need 512M no matter the OS. ?I think they >> will need all new PC's if they go that route. >> >> I am on the cusp of losing my network to Windows, at which point I will >> resign my post as unpaid volunteer sysadmin, so if I sound a little >> desperate, now you know why. >> >> Any help here would be greatly appreciated. >> >> -- >> Jim Thomas ? ? ? ? ? ?Principal Applications Engineer ?Bittware, Inc >> jthomas at bittware.com ?http://www.bittware.com ? ?(603) 226-0404 x536 >> The problem with the future is that it keeps turning into the present - >> Hobbes >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see > > From ltsp at symbio-technologies.com Tue Sep 20 20:36:13 2011 From: ltsp at symbio-technologies.com (Gideon Romm) Date: Tue, 20 Sep 2011 16:36:13 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E78E6F0.9090207@jomegat.com> References: <4E78E6F0.9090207@jomegat.com> Message-ID: Try enabling NBD_SWAP. If the clients have insufficient RAM, the best you can do is add swap space and hope the OS will swap out something else. -Gadi On Tue, Sep 20, 2011 at 3:18 PM, Jomegat wrote: > I have a K12Linux setup based on EL6-64. ?The TC's are ancient i686's with > 256M RAM and 10/100 nics. ?The TS has a gigE NIC, and between the TS & TCs > is a managed switch with 2 gigE ports and 24 10/100 ports. ?The TS is > connected via one of the gigE's. > > We have a student who is having real performance problems when he loads a > page. ?I went in today to be there while he was experiencing woe, and ran > the system monitor applet. > > When he loads the web page - with no flash in it that I can find - the > network maxes out at 12MiB/sec, and it takes an eternity for the page to > load. ?He reports that sometimes this particular page will cause the TC to > crash (black screen with text followed by login screen), but he was not able > to reproduce that while I was there. ?I suspect there is some user error > involved in the crash scenario, but it will be difficult to prove. > > I had tried running firefox as a local app last month, but the school needs > flash, and it requires 512M. ?Performance on these TC's was substantially > worse when running FF as a local app, so I backed off of that. > > The page he was trying to load is password protected, but I had him ctrl-U > while he had it open, and save the source to a file which I can share if > need be. ?It is filled with javascript, but I'm not sure that's the culprit > (though it seems the most likely suspect). ?I found no .flv, FLV, .swf, or > .SWF in the file anywhere. > > As an experiment, I connected a monitor, keybd, and mouse directly to the > server and had him login there. ?The page loads just fine that way. ?That > might be a possible solution, but for some reason, the display is incredibly > blurry. ?If I make that permanent, I'd have to run some cable through the > wall (they're pinched in the door right now, and barely reach). > > Before they called me in to look at this, he had been bringing his mom's > laptop to school so he could do his work (it's an online course). ?If it > crashes while he is taking a quiz, he loses a test attempt or gets a bad > grade, so this has to be reliable. > > His mom is on the school board, and she is pushing the school to abandon > LTSP in favor of a Windows solution. ?They think that if they install some > flavor of windows on the TC's they will get better performance, but I > believe they won't, as flash will need 512M no matter the OS. ?I think they > will need all new PC's if they go that route. > > I am on the cusp of losing my network to Windows, at which point I will > resign my post as unpaid volunteer sysadmin, so if I sound a little > desperate, now you know why. > > Any help here would be greatly appreciated. > > -- > Jim Thomas ? ? ? ? ? ?Principal Applications Engineer ?Bittware, Inc > jthomas at bittware.com ?http://www.bittware.com ? ?(603) 226-0404 x536 > The problem with the future is that it keeps turning into the present - > Hobbes > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > From jomegat at jomegat.com Tue Sep 20 20:51:18 2011 From: jomegat at jomegat.com (Jomegat) Date: Tue, 20 Sep 2011 16:51:18 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: References: <4E78E6F0.9090207@jomegat.com> Message-ID: <4E78FCC6.4060001@jomegat.com> On 09/20/2011 04:36 PM, Gideon Romm wrote: > Try enabling NBD_SWAP. If the clients have insufficient RAM, the best > you can do is add swap space and hope the OS will swap out something > else. Thanks. I've already got that enabled as best I can tell. Here's a snippet from chkconfig --list: xinetd based services: [snip] nbdrootd: on nbdswapd: on Maybe that's where the 21MiB/sec is coming from. As soon as I can get back to the school, I'll run top on the TC and see what's going on. Would it be possible (or advisable) to run swap on the TC from a local HD instead of over the network? > > -Gadi > > On Tue, Sep 20, 2011 at 3:18 PM, Jomegat wrote: >> I have a K12Linux setup based on EL6-64. The TC's are ancient i686's with >> 256M RAM and 10/100 nics. The TS has a gigE NIC, and between the TS& TCs >> is a managed switch with 2 gigE ports and 24 10/100 ports. The TS is >> connected via one of the gigE's. >> >> We have a student who is having real performance problems when he loads a >> page. I went in today to be there while he was experiencing woe, and ran >> the system monitor applet. >> >> When he loads the web page - with no flash in it that I can find - the >> network maxes out at 12MiB/sec, and it takes an eternity for the page to >> load. He reports that sometimes this particular page will cause the TC to >> crash (black screen with text followed by login screen), but he was not able >> to reproduce that while I was there. I suspect there is some user error >> involved in the crash scenario, but it will be difficult to prove. >> >> I had tried running firefox as a local app last month, but the school needs >> flash, and it requires 512M. Performance on these TC's was substantially >> worse when running FF as a local app, so I backed off of that. >> >> The page he was trying to load is password protected, but I had him ctrl-U >> while he had it open, and save the source to a file which I can share if >> need be. It is filled with javascript, but I'm not sure that's the culprit >> (though it seems the most likely suspect). I found no .flv, FLV, .swf, or >> .SWF in the file anywhere. >> >> As an experiment, I connected a monitor, keybd, and mouse directly to the >> server and had him login there. The page loads just fine that way. That >> might be a possible solution, but for some reason, the display is incredibly >> blurry. If I make that permanent, I'd have to run some cable through the >> wall (they're pinched in the door right now, and barely reach). >> >> Before they called me in to look at this, he had been bringing his mom's >> laptop to school so he could do his work (it's an online course). If it >> crashes while he is taking a quiz, he loses a test attempt or gets a bad >> grade, so this has to be reliable. >> >> His mom is on the school board, and she is pushing the school to abandon >> LTSP in favor of a Windows solution. They think that if they install some >> flavor of windows on the TC's they will get better performance, but I >> believe they won't, as flash will need 512M no matter the OS. I think they >> will need all new PC's if they go that route. >> >> I am on the cusp of losing my network to Windows, at which point I will >> resign my post as unpaid volunteer sysadmin, so if I sound a little >> desperate, now you know why. >> >> Any help here would be greatly appreciated. >> >> -- >> Jim Thomas Principal Applications Engineer Bittware, Inc >> jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 >> The problem with the future is that it keeps turning into the present - >> Hobbes >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see >> > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > -- Jim Thomas Principal Applications Engineer Bittware, Inc jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 The problem with the future is that it keeps turning into the present - Hobbes From jomegat at jomegat.com Tue Sep 20 20:57:42 2011 From: jomegat at jomegat.com (Jomegat) Date: Tue, 20 Sep 2011 16:57:42 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: References: <4E78E6F0.9090207@jomegat.com> Message-ID: <4E78FE46.7020701@jomegat.com> On 09/20/2011 04:34 PM, Dean Jones wrote: > +1 on the 64 bit flash. Much better performance if the problem is > flash related. The page he's trying to load has no flash. When I was testing, I had all the TC's playing full-screen youtube vids. It was a little laggy, but it did not fall to its knees. > As you know Windows will perform poorly on those clients as well > regardless of the server specs. It is dated hardware with little > video memory. I know that full well. I doubt the ability of the school board to understand that though until they give it a whirl. The TC's actually have HD's in them with Win2K. I've considered booting into that and giving them a demo. To add injury to insult, the board member I mentioned previously is bringing in a "computer person" to "evaluate our needs". Grrrr. Nevermind that I'm a software engineer with over 25 years of experience. > You can try nbdroot for a slight speed increase over NFS if the > problem is file i/o related. It's already using nbd. Thanks for your suggestions. > On Tue, Sep 20, 2011 at 1:12 PM, David Hopkins wrote: >> So, runs fine at server console, but lags severely at the thin client with >> the same account. Are you certain that the clients are running at 100Mb >> full duplex? If you've enabled a shell at the client can you check to see >> what ifconfig reports at the client? I would check memory/performance at >> the client with top (in a shell, add SCREEN_02=shell in lts.conf, ctl-alt-f2 >> to access it at the client) while the page is being displayed from the GUI >> login session just to convince myself it isn't a load issue at the client. >> >> Also, Adobe has a 64bit Flash for linux (v11) available as beta. I've been >> running that for the last couple of weeks (since it was released IOW) on my >> systems (Ubuntu 10.04 with LTSP packages added). >> http://labs.adobe.com/downloads/flashplayer11.html It definitely runs better >> than the 32bit flash with nswrapper. Installation was just a matter of >> uninstalling the 32bit flash and copying the 64bit version to the correct >> plugins directory. >> >> Those are the things that I'd check first. >> >> Sincerely, >> Dave Hopkins >> Newark Charter School >> >> On Tue, Sep 20, 2011 at 3:18 PM, Jomegat wrote: >>> >>> I have a K12Linux setup based on EL6-64. The TC's are ancient i686's with >>> 256M RAM and 10/100 nics. The TS has a gigE NIC, and between the TS& TCs >>> is a managed switch with 2 gigE ports and 24 10/100 ports. The TS is >>> connected via one of the gigE's. >>> >>> We have a student who is having real performance problems when he loads a >>> page. I went in today to be there while he was experiencing woe, and ran >>> the system monitor applet. >>> >>> When he loads the web page - with no flash in it that I can find - the >>> network maxes out at 12MiB/sec, and it takes an eternity for the page to >>> load. He reports that sometimes this particular page will cause the TC to >>> crash (black screen with text followed by login screen), but he was not able >>> to reproduce that while I was there. I suspect there is some user error >>> involved in the crash scenario, but it will be difficult to prove. >>> >>> I had tried running firefox as a local app last month, but the school >>> needs flash, and it requires 512M. Performance on these TC's was >>> substantially worse when running FF as a local app, so I backed off of that. >>> >>> The page he was trying to load is password protected, but I had him ctrl-U >>> while he had it open, and save the source to a file which I can share if >>> need be. It is filled with javascript, but I'm not sure that's the culprit >>> (though it seems the most likely suspect). I found no .flv, FLV, .swf, or >>> .SWF in the file anywhere. >>> >>> As an experiment, I connected a monitor, keybd, and mouse directly to the >>> server and had him login there. The page loads just fine that way. That >>> might be a possible solution, but for some reason, the display is incredibly >>> blurry. If I make that permanent, I'd have to run some cable through the >>> wall (they're pinched in the door right now, and barely reach). >>> >>> Before they called me in to look at this, he had been bringing his mom's >>> laptop to school so he could do his work (it's an online course). If it >>> crashes while he is taking a quiz, he loses a test attempt or gets a bad >>> grade, so this has to be reliable. >>> >>> His mom is on the school board, and she is pushing the school to abandon >>> LTSP in favor of a Windows solution. They think that if they install some >>> flavor of windows on the TC's they will get better performance, but I >>> believe they won't, as flash will need 512M no matter the OS. I think they >>> will need all new PC's if they go that route. >>> >>> I am on the cusp of losing my network to Windows, at which point I will >>> resign my post as unpaid volunteer sysadmin, so if I sound a little >>> desperate, now you know why. >>> >>> Any help here would be greatly appreciated. >>> >>> -- >>> Jim Thomas Principal Applications Engineer Bittware, Inc >>> jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 >>> The problem with the future is that it keeps turning into the present - >>> Hobbes >>> >>> _______________________________________________ >>> K12OSN mailing list >>> K12OSN at redhat.com >>> https://www.redhat.com/mailman/listinfo/k12osn >>> For more info see >> >> > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > -- Jim Thomas Principal Applications Engineer Bittware, Inc jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 The problem with the future is that it keeps turning into the present - Hobbes From ltsp at symbio-technologies.com Tue Sep 20 21:02:36 2011 From: ltsp at symbio-technologies.com (Gideon Romm) Date: Tue, 20 Sep 2011 17:02:36 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E78FCC6.4060001@jomegat.com> References: <4E78E6F0.9090207@jomegat.com> <4E78FCC6.4060001@jomegat.com> Message-ID: That just tells you the service is on. If you don't have NBD_SWAP in your lts.conf, then you are not using it. I believe there is also a parameter to use local swap from a hard drive, though I don't recall. Perhaps check the docs. Otherwise, if I remember, I will look in the code. -Gadi On Tue, Sep 20, 2011 at 4:51 PM, Jomegat wrote: > On 09/20/2011 04:36 PM, Gideon Romm wrote: >> >> Try enabling NBD_SWAP. ?If the clients have insufficient RAM, the best >> you can do is add swap space and hope the OS will swap out something >> else. > > Thanks. ?I've already got that enabled as best I can tell. ?Here's a snippet > from chkconfig --list: > > xinetd based services: > ?[snip] > ? ? ? ?nbdrootd: ? ? ? on > ? ? ? ?nbdswapd: ? ? ? on > > Maybe that's where the 21MiB/sec is coming from. > > As soon as I can get back to the school, I'll run top on the TC and see > what's going on. > > Would it be possible (or advisable) to run swap on the TC from a local HD > instead of over the network? > >> >> -Gadi >> >> On Tue, Sep 20, 2011 at 3:18 PM, Jomegat ?wrote: >>> >>> I have a K12Linux setup based on EL6-64. ?The TC's are ancient i686's >>> with >>> 256M RAM and 10/100 nics. ?The TS has a gigE NIC, and between the TS& >>> ?TCs >>> is a managed switch with 2 gigE ports and 24 10/100 ports. ?The TS is >>> connected via one of the gigE's. >>> >>> We have a student who is having real performance problems when he loads a >>> page. ?I went in today to be there while he was experiencing woe, and ran >>> the system monitor applet. >>> >>> When he loads the web page - with no flash in it that I can find - the >>> network maxes out at 12MiB/sec, and it takes an eternity for the page to >>> load. ?He reports that sometimes this particular page will cause the TC >>> to >>> crash (black screen with text followed by login screen), but he was not >>> able >>> to reproduce that while I was there. ?I suspect there is some user error >>> involved in the crash scenario, but it will be difficult to prove. >>> >>> I had tried running firefox as a local app last month, but the school >>> needs >>> flash, and it requires 512M. ?Performance on these TC's was substantially >>> worse when running FF as a local app, so I backed off of that. >>> >>> The page he was trying to load is password protected, but I had him >>> ctrl-U >>> while he had it open, and save the source to a file which I can share if >>> need be. ?It is filled with javascript, but I'm not sure that's the >>> culprit >>> (though it seems the most likely suspect). ?I found no .flv, FLV, .swf, >>> or >>> .SWF in the file anywhere. >>> >>> As an experiment, I connected a monitor, keybd, and mouse directly to the >>> server and had him login there. ?The page loads just fine that way. ?That >>> might be a possible solution, but for some reason, the display is >>> incredibly >>> blurry. ?If I make that permanent, I'd have to run some cable through the >>> wall (they're pinched in the door right now, and barely reach). >>> >>> Before they called me in to look at this, he had been bringing his mom's >>> laptop to school so he could do his work (it's an online course). ?If it >>> crashes while he is taking a quiz, he loses a test attempt or gets a bad >>> grade, so this has to be reliable. >>> >>> His mom is on the school board, and she is pushing the school to abandon >>> LTSP in favor of a Windows solution. ?They think that if they install >>> some >>> flavor of windows on the TC's they will get better performance, but I >>> believe they won't, as flash will need 512M no matter the OS. ?I think >>> they >>> will need all new PC's if they go that route. >>> >>> I am on the cusp of losing my network to Windows, at which point I will >>> resign my post as unpaid volunteer sysadmin, so if I sound a little >>> desperate, now you know why. >>> >>> Any help here would be greatly appreciated. >>> >>> -- >>> Jim Thomas ? ? ? ? ? ?Principal Applications Engineer ?Bittware, Inc >>> jthomas at bittware.com ?http://www.bittware.com ? ?(603) 226-0404 x536 >>> The problem with the future is that it keeps turning into the present - >>> Hobbes >>> >>> _______________________________________________ >>> K12OSN mailing list >>> K12OSN at redhat.com >>> https://www.redhat.com/mailman/listinfo/k12osn >>> For more info see >>> >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see >> > > > -- > Jim Thomas ? ? ? ? ? ?Principal Applications Engineer ?Bittware, Inc > jthomas at bittware.com ?http://www.bittware.com ? ?(603) 226-0404 x536 > The problem with the future is that it keeps turning into the present - > Hobbes > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > From ltsp at symbio-technologies.com Tue Sep 20 21:15:15 2011 From: ltsp at symbio-technologies.com (Gideon Romm) Date: Tue, 20 Sep 2011 17:15:15 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: References: <4E78E6F0.9090207@jomegat.com> <4E78FCC6.4060001@jomegat.com> Message-ID: Keep in mind that although apps run on the server, they still make use of Xserver memory to load images. Xserver memory is on the client. It is very easy to have apps crash a thin client because they overload the available memory of the Xserver on the client. Swap can help mitigate this. Also, you can try booting with the "nocompcache" kernel parameter (added to your pxelinux.cfg/default file). It is debatable as to whether the default use of compcache helps or hinders on low ram clients. -Gadi On Tue, Sep 20, 2011 at 5:02 PM, Gideon Romm wrote: > That just tells you the service is on. If you don't have NBD_SWAP in > your lts.conf, then you are not using it. > > I believe there is also a parameter to use local swap from a hard > drive, though I don't recall. Perhaps check the docs. Otherwise, if I > remember, I will look in the code. > > -Gadi > > On Tue, Sep 20, 2011 at 4:51 PM, Jomegat wrote: >> On 09/20/2011 04:36 PM, Gideon Romm wrote: >>> >>> Try enabling NBD_SWAP. ?If the clients have insufficient RAM, the best >>> you can do is add swap space and hope the OS will swap out something >>> else. >> >> Thanks. ?I've already got that enabled as best I can tell. ?Here's a snippet >> from chkconfig --list: >> >> xinetd based services: >> ?[snip] >> ? ? ? ?nbdrootd: ? ? ? on >> ? ? ? ?nbdswapd: ? ? ? on >> >> Maybe that's where the 21MiB/sec is coming from. >> >> As soon as I can get back to the school, I'll run top on the TC and see >> what's going on. >> >> Would it be possible (or advisable) to run swap on the TC from a local HD >> instead of over the network? >> >>> >>> -Gadi >>> >>> On Tue, Sep 20, 2011 at 3:18 PM, Jomegat ?wrote: >>>> >>>> I have a K12Linux setup based on EL6-64. ?The TC's are ancient i686's >>>> with >>>> 256M RAM and 10/100 nics. ?The TS has a gigE NIC, and between the TS& >>>> ?TCs >>>> is a managed switch with 2 gigE ports and 24 10/100 ports. ?The TS is >>>> connected via one of the gigE's. >>>> >>>> We have a student who is having real performance problems when he loads a >>>> page. ?I went in today to be there while he was experiencing woe, and ran >>>> the system monitor applet. >>>> >>>> When he loads the web page - with no flash in it that I can find - the >>>> network maxes out at 12MiB/sec, and it takes an eternity for the page to >>>> load. ?He reports that sometimes this particular page will cause the TC >>>> to >>>> crash (black screen with text followed by login screen), but he was not >>>> able >>>> to reproduce that while I was there. ?I suspect there is some user error >>>> involved in the crash scenario, but it will be difficult to prove. >>>> >>>> I had tried running firefox as a local app last month, but the school >>>> needs >>>> flash, and it requires 512M. ?Performance on these TC's was substantially >>>> worse when running FF as a local app, so I backed off of that. >>>> >>>> The page he was trying to load is password protected, but I had him >>>> ctrl-U >>>> while he had it open, and save the source to a file which I can share if >>>> need be. ?It is filled with javascript, but I'm not sure that's the >>>> culprit >>>> (though it seems the most likely suspect). ?I found no .flv, FLV, .swf, >>>> or >>>> .SWF in the file anywhere. >>>> >>>> As an experiment, I connected a monitor, keybd, and mouse directly to the >>>> server and had him login there. ?The page loads just fine that way. ?That >>>> might be a possible solution, but for some reason, the display is >>>> incredibly >>>> blurry. ?If I make that permanent, I'd have to run some cable through the >>>> wall (they're pinched in the door right now, and barely reach). >>>> >>>> Before they called me in to look at this, he had been bringing his mom's >>>> laptop to school so he could do his work (it's an online course). ?If it >>>> crashes while he is taking a quiz, he loses a test attempt or gets a bad >>>> grade, so this has to be reliable. >>>> >>>> His mom is on the school board, and she is pushing the school to abandon >>>> LTSP in favor of a Windows solution. ?They think that if they install >>>> some >>>> flavor of windows on the TC's they will get better performance, but I >>>> believe they won't, as flash will need 512M no matter the OS. ?I think >>>> they >>>> will need all new PC's if they go that route. >>>> >>>> I am on the cusp of losing my network to Windows, at which point I will >>>> resign my post as unpaid volunteer sysadmin, so if I sound a little >>>> desperate, now you know why. >>>> >>>> Any help here would be greatly appreciated. >>>> >>>> -- >>>> Jim Thomas ? ? ? ? ? ?Principal Applications Engineer ?Bittware, Inc >>>> jthomas at bittware.com ?http://www.bittware.com ? ?(603) 226-0404 x536 >>>> The problem with the future is that it keeps turning into the present - >>>> Hobbes >>>> >>>> _______________________________________________ >>>> K12OSN mailing list >>>> K12OSN at redhat.com >>>> https://www.redhat.com/mailman/listinfo/k12osn >>>> For more info see >>>> >>> >>> _______________________________________________ >>> K12OSN mailing list >>> K12OSN at redhat.com >>> https://www.redhat.com/mailman/listinfo/k12osn >>> For more info see >>> >> >> >> -- >> Jim Thomas ? ? ? ? ? ?Principal Applications Engineer ?Bittware, Inc >> jthomas at bittware.com ?http://www.bittware.com ? ?(603) 226-0404 x536 >> The problem with the future is that it keeps turning into the present - >> Hobbes >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see >> > From news at siddall.name Tue Sep 20 22:13:57 2011 From: news at siddall.name (Jeff Siddall) Date: Tue, 20 Sep 2011 18:13:57 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E78E6F0.9090207@jomegat.com> References: <4E78E6F0.9090207@jomegat.com> Message-ID: <4E791025.30001@siddall.name> On 09/20/2011 03:18 PM, Jomegat wrote: > When he loads the web page - with no flash in it that I can find - the > network maxes out at 12MiB/sec, and it takes an eternity for the page to > load. He reports that sometimes this particular page will cause the TC > to crash (black screen with text followed by login screen), but he was > not able to reproduce that while I was there. I suspect there is some > user error involved in the crash scenario, but it will be difficult to > prove. Unfortunately there are some cases where LTSP just doesn't work well. Sounds like you hit one of those where javascript is doing something silly with the display and sending raw display updates across the LAN is killing things. You can probably get around the problem of that one page by running the browser as a localapp, but of course then there is no flash for that user. That being said, there are a few things you can do: 1. Try a different browser (ex: chrome). It may work better and it would allow you to run a dual browser (ex: firefox as a localapp and chrome as a native). If the users will put up with it tell them to run flash sites in Chrome and all others in Firefox. Ugly for sure but may work for the few exception cases you are running into. 2. Fire up one of those clients as a Windoze machine and load a flash page to show that Windows on your clients is a non-option. 3. Try installing more RAM in one client and see if it works any better on the problem page. That should confirm/deny the running out of RAM question. Also, that will allow you to try installing flash in the chroot and see if the performance is acceptable. Maybe you can add some RAM to your clients and successfully run them all with localapp browsers. 4. If they are going to upgrade the PCs anyway, try to convince them to order them diskless initially and run either LTSP with localapps or something like DRBL. If nothing else they save admin time/effort vs. any Windows option plus the cost of all the hard drives and Windoze licenses in the clients. They can still add drives and Windoze after if it doesn't work. Good luck! Jeff From jomegat at jomegat.com Tue Sep 20 22:51:50 2011 From: jomegat at jomegat.com (Jomegat) Date: Tue, 20 Sep 2011 18:51:50 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: References: <4E78E6F0.9090207@jomegat.com> <4E78FCC6.4060001@jomegat.com> Message-ID: <4E791906.20402@jomegat.com> On 09/20/2011 05:02 PM, Gideon Romm wrote: > That just tells you the service is on. If you don't have NBD_SWAP in > your lts.conf, then you are not using it. OK, there was nothing of the sort in lts.conf. I added this line: NBD_SWAP=on I'll test it when I can get over there. Thanks a million. > > I believe there is also a parameter to use local swap from a hard > drive, though I don't recall. Perhaps check the docs. Otherwise, if I > remember, I will look in the code. > > -Gadi > > On Tue, Sep 20, 2011 at 4:51 PM, Jomegat wrote: >> On 09/20/2011 04:36 PM, Gideon Romm wrote: >>> >>> Try enabling NBD_SWAP. If the clients have insufficient RAM, the best >>> you can do is add swap space and hope the OS will swap out something >>> else. >> >> Thanks. I've already got that enabled as best I can tell. Here's a snippet >> from chkconfig --list: >> >> xinetd based services: >> [snip] >> nbdrootd: on >> nbdswapd: on >> >> Maybe that's where the 21MiB/sec is coming from. >> >> As soon as I can get back to the school, I'll run top on the TC and see >> what's going on. >> >> Would it be possible (or advisable) to run swap on the TC from a local HD >> instead of over the network? >> >>> >>> -Gadi >>> >>> On Tue, Sep 20, 2011 at 3:18 PM, Jomegat wrote: >>>> >>>> I have a K12Linux setup based on EL6-64. The TC's are ancient i686's >>>> with >>>> 256M RAM and 10/100 nics. The TS has a gigE NIC, and between the TS& >>>> TCs >>>> is a managed switch with 2 gigE ports and 24 10/100 ports. The TS is >>>> connected via one of the gigE's. >>>> >>>> We have a student who is having real performance problems when he loads a >>>> page. I went in today to be there while he was experiencing woe, and ran >>>> the system monitor applet. >>>> >>>> When he loads the web page - with no flash in it that I can find - the >>>> network maxes out at 12MiB/sec, and it takes an eternity for the page to >>>> load. He reports that sometimes this particular page will cause the TC >>>> to >>>> crash (black screen with text followed by login screen), but he was not >>>> able >>>> to reproduce that while I was there. I suspect there is some user error >>>> involved in the crash scenario, but it will be difficult to prove. >>>> >>>> I had tried running firefox as a local app last month, but the school >>>> needs >>>> flash, and it requires 512M. Performance on these TC's was substantially >>>> worse when running FF as a local app, so I backed off of that. >>>> >>>> The page he was trying to load is password protected, but I had him >>>> ctrl-U >>>> while he had it open, and save the source to a file which I can share if >>>> need be. It is filled with javascript, but I'm not sure that's the >>>> culprit >>>> (though it seems the most likely suspect). I found no .flv, FLV, .swf, >>>> or >>>> .SWF in the file anywhere. >>>> >>>> As an experiment, I connected a monitor, keybd, and mouse directly to the >>>> server and had him login there. The page loads just fine that way. That >>>> might be a possible solution, but for some reason, the display is >>>> incredibly >>>> blurry. If I make that permanent, I'd have to run some cable through the >>>> wall (they're pinched in the door right now, and barely reach). >>>> >>>> Before they called me in to look at this, he had been bringing his mom's >>>> laptop to school so he could do his work (it's an online course). If it >>>> crashes while he is taking a quiz, he loses a test attempt or gets a bad >>>> grade, so this has to be reliable. >>>> >>>> His mom is on the school board, and she is pushing the school to abandon >>>> LTSP in favor of a Windows solution. They think that if they install >>>> some >>>> flavor of windows on the TC's they will get better performance, but I >>>> believe they won't, as flash will need 512M no matter the OS. I think >>>> they >>>> will need all new PC's if they go that route. >>>> >>>> I am on the cusp of losing my network to Windows, at which point I will >>>> resign my post as unpaid volunteer sysadmin, so if I sound a little >>>> desperate, now you know why. >>>> >>>> Any help here would be greatly appreciated. >>>> >>>> -- >>>> Jim Thomas Principal Applications Engineer Bittware, Inc >>>> jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 >>>> The problem with the future is that it keeps turning into the present - >>>> Hobbes >>>> >>>> _______________________________________________ >>>> K12OSN mailing list >>>> K12OSN at redhat.com >>>> https://www.redhat.com/mailman/listinfo/k12osn >>>> For more info see >>>> >>> >>> _______________________________________________ >>> K12OSN mailing list >>> K12OSN at redhat.com >>> https://www.redhat.com/mailman/listinfo/k12osn >>> For more info see >>> >> >> >> -- >> Jim Thomas Principal Applications Engineer Bittware, Inc >> jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 >> The problem with the future is that it keeps turning into the present - >> Hobbes >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see >> > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > -- Jim Thomas jomegat at jomegat.com "And" is a word that should never be used at the beginning of a sentence. From jomegat at jomegat.com Tue Sep 20 23:00:06 2011 From: jomegat at jomegat.com (Jomegat) Date: Tue, 20 Sep 2011 19:00:06 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E791025.30001@siddall.name> References: <4E78E6F0.9090207@jomegat.com> <4E791025.30001@siddall.name> Message-ID: <4E791AF6.4070507@jomegat.com> On 09/20/2011 06:13 PM, Jeff Siddall wrote: > Unfortunately there are some cases where LTSP just doesn't work well. > Sounds like you hit one of those where javascript is doing something > silly with the display and sending raw display updates across the LAN is > killing things. Yes. I think I'm falling victim to bad programming here. > 2. Fire up one of those clients as a Windoze machine and load a flash > page to show that Windows on your clients is a non-option. I think I need to do that pretty soon. Anyone know what I have to do to my server to make it forward network access? That way I wouldn't have to move any cables. > 3. Try installing more RAM in one client and see if it works any better > on the problem page. That should confirm/deny the running out of RAM > question. Also, that will allow you to try installing flash in the > chroot and see if the performance is acceptable. Maybe you can add some > RAM to your clients and successfully run them all with localapp browsers. I'll see if I can scare up some RAM. > 4. If they are going to upgrade the PCs anyway, try to convince them to > order them diskless initially and run either LTSP with localapps or > something like DRBL. If nothing else they save admin time/effort vs. any > Windows option plus the cost of all the hard drives and Windoze licenses > in the clients. They can still add drives and Windoze after if it > doesn't work. That sounds like a good plan. The other thing they must understand is that they will lose a free sysadmin. I would not run a Windows network for money, much less for free. As a result, I have next to no Windows admin skills. My fear is that with grade-school aged kids surfing the web on a Windows machine, the network will be chock-full-o-malware in under a month. Then they get to hire someone to do janitorial work at an hourly rate. I ran a K12LTSP network for four years when I lived in Virginia, and of course, we were malware-free that whole time. This is our fourth year of LTSP at this school, and still no malware on my watch. The problem is going to be getting them to recognize that. -- Jim Thomas jomegat at jomegat.com "And" is a word that should never be used at the beginning of a sentence. From news at siddall.name Tue Sep 20 23:42:49 2011 From: news at siddall.name (Jeff Siddall) Date: Tue, 20 Sep 2011 19:42:49 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E791AF6.4070507@jomegat.com> References: <4E78E6F0.9090207@jomegat.com> <4E791025.30001@siddall.name> <4E791AF6.4070507@jomegat.com> Message-ID: <4E7924F9.5020902@siddall.name> On 09/20/2011 07:00 PM, Jomegat wrote: > I think I need to do that pretty soon. Anyone know what I have to do to > my server to make it forward network access? That way I wouldn't have to > move any cables. If you just want a temporary (ie: until the next reboot) routing between NICs you can just run: echo 1 > /proc/sys/net/ipv4/ip_forwarding as root. > That sounds like a good plan. The other thing they must understand is > that they will lose a free sysadmin. I would not run a Windows network > for money, much less for free. As a result, I have next to no Windows > admin skills. My fear is that with grade-school aged kids surfing the > web on a Windows machine, the network will be chock-full-o-malware in > under a month. Then they get to hire someone to do janitorial work at an > hourly rate. Yup. I volunteer running an LTSP server for a non-profit. Most people are on thin clients but a few special cases still run Windoze fat clients. They pay someone else for their Windoze support so they are very aware how unreliable and expensive the Windoze systems are. They have had no unplanned outages on the terminal server in the 2-1/2 years it has been running. Jeff From peter at scheie.homedns.org Tue Sep 20 23:10:11 2011 From: peter at scheie.homedns.org (Peter Scheie) Date: Tue, 20 Sep 2011 18:10:11 -0500 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E791AF6.4070507@jomegat.com> References: <4E78E6F0.9090207@jomegat.com> <4E791025.30001@siddall.name> <4E791AF6.4070507@jomegat.com> Message-ID: <4E791D53.9090907@scheie.homedns.org> Jomegat wrote: >> 2. Fire up one of those clients as a Windoze machine and load a flash >> page to show that Windows on your clients is a non-option. > > I think I need to do that pretty soon. Anyone know what I have to do to > my server to make it forward network access? That way I wouldn't have > to move any cables. > From a wiki page I wrote for myself a few years ago. It may be a bit dated. "Enabling fat clients to work on the thin client segment K12LTSP has this enabled by default; K12Linux does not. To enable it: * echo 1 >/proc/sys/net/ipv4/ip_forward * edit /etc/sysctl.conf to make that permanent * iptables -t nat -A POSTROUTING -j MASQUERADE * iptables-save * service dnsmasq start The first two turn on forwarding, which is the primary requirement. The next two tell iptables to allow port forwarding; if iptables is turned off, they are unnecessary. The last one is to provide a DNS service to the fat client. Remember that thin client's apps are running on the server so they use the DNS server listed in /etc/resolv.conf. On the TC segment, however, dhcpd.conf tells clients that the nameserver is the LTSP server, meaning there needs to be a nameserver running on the LTSP server. That's what dnsmasq does. Or you can give out the address of a different DNS server via dhcpd.conf." HTH Peter From ltsp at symbio-technologies.com Wed Sep 21 01:01:32 2011 From: ltsp at symbio-technologies.com (Gideon Romm) Date: Tue, 20 Sep 2011 21:01:32 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E791906.20402@jomegat.com> References: <4E78E6F0.9090207@jomegat.com> <4E78FCC6.4060001@jomegat.com> <4E791906.20402@jomegat.com> Message-ID: Should be NBD_SWAP = True (it won't understand "on") :) On Tue, Sep 20, 2011 at 6:51 PM, Jomegat wrote: > On 09/20/2011 05:02 PM, Gideon Romm wrote: >> >> That just tells you the service is on. If you don't have NBD_SWAP in >> your lts.conf, then you are not using it. > > OK, there was nothing of the sort in lts.conf. ?I added this line: > NBD_SWAP=on > > I'll test it when I can get over there. > > Thanks a million. > >> >> I believe there is also a parameter to use local swap from a hard >> drive, though I don't recall. Perhaps check the docs. Otherwise, if I >> remember, I will look in the code. >> >> -Gadi >> >> On Tue, Sep 20, 2011 at 4:51 PM, Jomegat ?wrote: >>> >>> On 09/20/2011 04:36 PM, Gideon Romm wrote: >>>> >>>> Try enabling NBD_SWAP. ?If the clients have insufficient RAM, the best >>>> you can do is add swap space and hope the OS will swap out something >>>> else. >>> >>> Thanks. ?I've already got that enabled as best I can tell. ?Here's a >>> snippet >>> from chkconfig --list: >>> >>> xinetd based services: >>> ?[snip] >>> ? ? ? ?nbdrootd: ? ? ? on >>> ? ? ? ?nbdswapd: ? ? ? on >>> >>> Maybe that's where the 21MiB/sec is coming from. >>> >>> As soon as I can get back to the school, I'll run top on the TC and see >>> what's going on. >>> >>> Would it be possible (or advisable) to run swap on the TC from a local HD >>> instead of over the network? >>> >>>> >>>> -Gadi >>>> >>>> On Tue, Sep 20, 2011 at 3:18 PM, Jomegat ? ?wrote: >>>>> >>>>> I have a K12Linux setup based on EL6-64. ?The TC's are ancient i686's >>>>> with >>>>> 256M RAM and 10/100 nics. ?The TS has a gigE NIC, and between the TS& >>>>> ?TCs >>>>> is a managed switch with 2 gigE ports and 24 10/100 ports. ?The TS is >>>>> connected via one of the gigE's. >>>>> >>>>> We have a student who is having real performance problems when he loads >>>>> a >>>>> page. ?I went in today to be there while he was experiencing woe, and >>>>> ran >>>>> the system monitor applet. >>>>> >>>>> When he loads the web page - with no flash in it that I can find - the >>>>> network maxes out at 12MiB/sec, and it takes an eternity for the page >>>>> to >>>>> load. ?He reports that sometimes this particular page will cause the TC >>>>> to >>>>> crash (black screen with text followed by login screen), but he was not >>>>> able >>>>> to reproduce that while I was there. ?I suspect there is some user >>>>> error >>>>> involved in the crash scenario, but it will be difficult to prove. >>>>> >>>>> I had tried running firefox as a local app last month, but the school >>>>> needs >>>>> flash, and it requires 512M. ?Performance on these TC's was >>>>> substantially >>>>> worse when running FF as a local app, so I backed off of that. >>>>> >>>>> The page he was trying to load is password protected, but I had him >>>>> ctrl-U >>>>> while he had it open, and save the source to a file which I can share >>>>> if >>>>> need be. ?It is filled with javascript, but I'm not sure that's the >>>>> culprit >>>>> (though it seems the most likely suspect). ?I found no .flv, FLV, .swf, >>>>> or >>>>> .SWF in the file anywhere. >>>>> >>>>> As an experiment, I connected a monitor, keybd, and mouse directly to >>>>> the >>>>> server and had him login there. ?The page loads just fine that way. >>>>> ?That >>>>> might be a possible solution, but for some reason, the display is >>>>> incredibly >>>>> blurry. ?If I make that permanent, I'd have to run some cable through >>>>> the >>>>> wall (they're pinched in the door right now, and barely reach). >>>>> >>>>> Before they called me in to look at this, he had been bringing his >>>>> mom's >>>>> laptop to school so he could do his work (it's an online course). ?If >>>>> it >>>>> crashes while he is taking a quiz, he loses a test attempt or gets a >>>>> bad >>>>> grade, so this has to be reliable. >>>>> >>>>> His mom is on the school board, and she is pushing the school to >>>>> abandon >>>>> LTSP in favor of a Windows solution. ?They think that if they install >>>>> some >>>>> flavor of windows on the TC's they will get better performance, but I >>>>> believe they won't, as flash will need 512M no matter the OS. ?I think >>>>> they >>>>> will need all new PC's if they go that route. >>>>> >>>>> I am on the cusp of losing my network to Windows, at which point I will >>>>> resign my post as unpaid volunteer sysadmin, so if I sound a little >>>>> desperate, now you know why. >>>>> >>>>> Any help here would be greatly appreciated. >>>>> >>>>> -- >>>>> Jim Thomas ? ? ? ? ? ?Principal Applications Engineer ?Bittware, Inc >>>>> jthomas at bittware.com ?http://www.bittware.com ? ?(603) 226-0404 x536 >>>>> The problem with the future is that it keeps turning into the present - >>>>> Hobbes >>>>> >>>>> _______________________________________________ >>>>> K12OSN mailing list >>>>> K12OSN at redhat.com >>>>> https://www.redhat.com/mailman/listinfo/k12osn >>>>> For more info see >>>>> >>>> >>>> _______________________________________________ >>>> K12OSN mailing list >>>> K12OSN at redhat.com >>>> https://www.redhat.com/mailman/listinfo/k12osn >>>> For more info see >>>> >>> >>> >>> -- >>> Jim Thomas ? ? ? ? ? ?Principal Applications Engineer ?Bittware, Inc >>> jthomas at bittware.com ?http://www.bittware.com ? ?(603) 226-0404 x536 >>> The problem with the future is that it keeps turning into the present - >>> Hobbes >>> >>> _______________________________________________ >>> K12OSN mailing list >>> K12OSN at redhat.com >>> https://www.redhat.com/mailman/listinfo/k12osn >>> For more info see >>> >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see >> > > > -- > Jim Thomas ? ? ? ? ? ?jomegat at jomegat.com > ?"And" is a word that should never be used at the beginning of a sentence. > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > From jomegat at jomegat.com Wed Sep 21 01:07:44 2011 From: jomegat at jomegat.com (Jomegat) Date: Tue, 20 Sep 2011 21:07:44 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: References: <4E78E6F0.9090207@jomegat.com> <4E78FCC6.4060001@jomegat.com> <4E791906.20402@jomegat.com> Message-ID: <4E7938E0.7050302@jomegat.com> On 09/20/2011 09:01 PM, Gideon Romm wrote: > Should be NBD_SWAP = True > > (it won't understand "on") :) Noted and applied. Thanks! > > On Tue, Sep 20, 2011 at 6:51 PM, Jomegat wrote: >> On 09/20/2011 05:02 PM, Gideon Romm wrote: >>> >>> That just tells you the service is on. If you don't have NBD_SWAP in >>> your lts.conf, then you are not using it. >> >> OK, there was nothing of the sort in lts.conf. I added this line: >> NBD_SWAP=on >> >> I'll test it when I can get over there. >> >> Thanks a million. >> >>> >>> I believe there is also a parameter to use local swap from a hard >>> drive, though I don't recall. Perhaps check the docs. Otherwise, if I >>> remember, I will look in the code. >>> >>> -Gadi >>> >>> On Tue, Sep 20, 2011 at 4:51 PM, Jomegat wrote: >>>> >>>> On 09/20/2011 04:36 PM, Gideon Romm wrote: >>>>> >>>>> Try enabling NBD_SWAP. If the clients have insufficient RAM, the best >>>>> you can do is add swap space and hope the OS will swap out something >>>>> else. >>>> >>>> Thanks. I've already got that enabled as best I can tell. Here's a >>>> snippet >>>> from chkconfig --list: >>>> >>>> xinetd based services: >>>> [snip] >>>> nbdrootd: on >>>> nbdswapd: on >>>> >>>> Maybe that's where the 21MiB/sec is coming from. >>>> >>>> As soon as I can get back to the school, I'll run top on the TC and see >>>> what's going on. >>>> >>>> Would it be possible (or advisable) to run swap on the TC from a local HD >>>> instead of over the network? >>>> >>>>> >>>>> -Gadi >>>>> >>>>> On Tue, Sep 20, 2011 at 3:18 PM, Jomegat wrote: >>>>>> >>>>>> I have a K12Linux setup based on EL6-64. The TC's are ancient i686's >>>>>> with >>>>>> 256M RAM and 10/100 nics. The TS has a gigE NIC, and between the TS& >>>>>> TCs >>>>>> is a managed switch with 2 gigE ports and 24 10/100 ports. The TS is >>>>>> connected via one of the gigE's. >>>>>> >>>>>> We have a student who is having real performance problems when he loads >>>>>> a >>>>>> page. I went in today to be there while he was experiencing woe, and >>>>>> ran >>>>>> the system monitor applet. >>>>>> >>>>>> When he loads the web page - with no flash in it that I can find - the >>>>>> network maxes out at 12MiB/sec, and it takes an eternity for the page >>>>>> to >>>>>> load. He reports that sometimes this particular page will cause the TC >>>>>> to >>>>>> crash (black screen with text followed by login screen), but he was not >>>>>> able >>>>>> to reproduce that while I was there. I suspect there is some user >>>>>> error >>>>>> involved in the crash scenario, but it will be difficult to prove. >>>>>> >>>>>> I had tried running firefox as a local app last month, but the school >>>>>> needs >>>>>> flash, and it requires 512M. Performance on these TC's was >>>>>> substantially >>>>>> worse when running FF as a local app, so I backed off of that. >>>>>> >>>>>> The page he was trying to load is password protected, but I had him >>>>>> ctrl-U >>>>>> while he had it open, and save the source to a file which I can share >>>>>> if >>>>>> need be. It is filled with javascript, but I'm not sure that's the >>>>>> culprit >>>>>> (though it seems the most likely suspect). I found no .flv, FLV, .swf, >>>>>> or >>>>>> .SWF in the file anywhere. >>>>>> >>>>>> As an experiment, I connected a monitor, keybd, and mouse directly to >>>>>> the >>>>>> server and had him login there. The page loads just fine that way. >>>>>> That >>>>>> might be a possible solution, but for some reason, the display is >>>>>> incredibly >>>>>> blurry. If I make that permanent, I'd have to run some cable through >>>>>> the >>>>>> wall (they're pinched in the door right now, and barely reach). >>>>>> >>>>>> Before they called me in to look at this, he had been bringing his >>>>>> mom's >>>>>> laptop to school so he could do his work (it's an online course). If >>>>>> it >>>>>> crashes while he is taking a quiz, he loses a test attempt or gets a >>>>>> bad >>>>>> grade, so this has to be reliable. >>>>>> >>>>>> His mom is on the school board, and she is pushing the school to >>>>>> abandon >>>>>> LTSP in favor of a Windows solution. They think that if they install >>>>>> some >>>>>> flavor of windows on the TC's they will get better performance, but I >>>>>> believe they won't, as flash will need 512M no matter the OS. I think >>>>>> they >>>>>> will need all new PC's if they go that route. >>>>>> >>>>>> I am on the cusp of losing my network to Windows, at which point I will >>>>>> resign my post as unpaid volunteer sysadmin, so if I sound a little >>>>>> desperate, now you know why. >>>>>> >>>>>> Any help here would be greatly appreciated. >>>>>> >>>>>> -- >>>>>> Jim Thomas Principal Applications Engineer Bittware, Inc >>>>>> jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 >>>>>> The problem with the future is that it keeps turning into the present - >>>>>> Hobbes >>>>>> >>>>>> _______________________________________________ >>>>>> K12OSN mailing list >>>>>> K12OSN at redhat.com >>>>>> https://www.redhat.com/mailman/listinfo/k12osn >>>>>> For more info see >>>>>> >>>>> >>>>> _______________________________________________ >>>>> K12OSN mailing list >>>>> K12OSN at redhat.com >>>>> https://www.redhat.com/mailman/listinfo/k12osn >>>>> For more info see >>>>> >>>> >>>> >>>> -- >>>> Jim Thomas Principal Applications Engineer Bittware, Inc >>>> jthomas at bittware.com http://www.bittware.com (603) 226-0404 x536 >>>> The problem with the future is that it keeps turning into the present - >>>> Hobbes >>>> >>>> _______________________________________________ >>>> K12OSN mailing list >>>> K12OSN at redhat.com >>>> https://www.redhat.com/mailman/listinfo/k12osn >>>> For more info see >>>> >>> >>> _______________________________________________ >>> K12OSN mailing list >>> K12OSN at redhat.com >>> https://www.redhat.com/mailman/listinfo/k12osn >>> For more info see >>> >> >> >> -- >> Jim Thomas jomegat at jomegat.com >> "And" is a word that should never be used at the beginning of a sentence. >> >> _______________________________________________ >> K12OSN mailing list >> K12OSN at redhat.com >> https://www.redhat.com/mailman/listinfo/k12osn >> For more info see >> > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > -- Jim Thomas jomegat at jomegat.com "And" is a word that should never be used at the beginning of a sentence. From carl at snarlnet.com Wed Sep 21 01:34:56 2011 From: carl at snarlnet.com (Carl Keil) Date: Tue, 20 Sep 2011 18:34:56 -0700 Subject: [K12OSN] Poor performance woes Message-ID: <4E793F40.6010505@snarlnet.com> I just keep thinking that no system can really be all things to all people, no matter how good it is. Like that office mentioned that has a few Windows machines. Could you scare up the resources to put another NIC in the server and have a few "higher-powered" thin clients that have more ram, dedicated Gig-E to the new NIC on the server and better video cards, for the kids whose parents are on the school board? I'm serious, aside from that last bit of snarkiness, sometimes you just need a better workstation, even if 99% of the time you don't. Maybe run a few, selected local apps on those. ck From jomegat at jomegat.com Wed Sep 21 01:40:40 2011 From: jomegat at jomegat.com (Jomegat) Date: Tue, 20 Sep 2011 21:40:40 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E793F40.6010505@snarlnet.com> References: <4E793F40.6010505@snarlnet.com> Message-ID: <4E794098.5030405@jomegat.com> On 09/20/2011 09:34 PM, Carl Keil wrote: > I just keep thinking that no system can really be all things to all > people, no matter how good it is. Like that office mentioned that has a > few Windows machines. Could you scare up the resources to put another > NIC in the server and have a few "higher-powered" thin clients that have > more ram, dedicated Gig-E to the new NIC on the server and better video > cards, for the kids whose parents are on the school board? I'm serious, > aside from that last bit of snarkiness, sometimes you just need a better > workstation, even if 99% of the time you don't. Maybe run a few, > selected local apps on those. That's not a bad idea. I do have a spare PC that's at least five years newer than my TC's. I should try one. -- Jim Thomas jomegat at jomegat.com "And" is a word that should never be used at the beginning of a sentence. From carl at snarlnet.com Wed Sep 21 16:34:45 2011 From: carl at snarlnet.com (Carl Keil) Date: Wed, 21 Sep 2011 09:34:45 -0700 Subject: [K12OSN] Poor performance woes Message-ID: <4E7A1225.7070609@snarlnet.com> > Could you scare up the resources to put another > NIC in the server and have a few "higher-powered" thin clients that have > more ram, dedicated Gig-E to the new NIC on the server and better video cards, for the kids whose parents are on the school board? > That's not a bad idea. I do have a spare PC that's at least five years > newer than my TC's. I should try one. I was thinking, from the discussion, that your bottleneck is probably ethernet speed, video card or RAM on the client. By "higher powered" I just meant improved in those areas. Might not even need full-on new systems, aside from the local apps angle. From lesmikesell at gmail.com Wed Sep 21 16:41:43 2011 From: lesmikesell at gmail.com (Les Mikesell) Date: Wed, 21 Sep 2011 11:41:43 -0500 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E7A1225.7070609@snarlnet.com> References: <4E7A1225.7070609@snarlnet.com> Message-ID: On Wed, Sep 21, 2011 at 11:34 AM, Carl Keil wrote: > >> Could you scare up the resources to put another >> NIC in the server and have a few "higher-powered" thin clients that have >> more ram, dedicated Gig-E to the new NIC on the server and better video >> cards, for the kids whose parents are on the school board? ?That's not >> a bad idea. ?I do have a spare PC that's at least five years >> newer than my TC's. ?I should try one. > > I was thinking, from the discussion, that your bottleneck is probably > ethernet speed, video card or RAM on the client. ?By "higher powered" I just > meant improved in those areas. ?Might not even need full-on new systems, > aside from the local apps angle. And to approach the problem from a slightly different angle, you might try using freenx on the server with the NX client on the client as a local app (maybe test with a windows laptop first to see if it looks promising). That is very similar to normal thin-client operation in that the desktop session is running on the server but the proxy/caching/compression layers might cover up your bottleneck. Having freenx installed shouldn't interfere with normal ltsp use and it also permits easy access without rebooting from windows/mac clients. -- Les Mikesell lesmikesell at gmail.com From tom.hoffman at gmail.com Wed Sep 21 17:27:18 2011 From: tom.hoffman at gmail.com (Tom Hoffman) Date: Wed, 21 Sep 2011 13:27:18 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: References: <4E7A1225.7070609@snarlnet.com> Message-ID: This might not solve your problem, but I know Chromium at least ships with some pretty sophisticated developer tools for profiling and analyzing javascript and browser performance. If you poke around a bit you might get an idea of what's causing this weird behavior. I know I'm curious... --Tom On Wed, Sep 21, 2011 at 12:41 PM, Les Mikesell wrote: > On Wed, Sep 21, 2011 at 11:34 AM, Carl Keil wrote: >> >>> Could you scare up the resources to put another >>> NIC in the server and have a few "higher-powered" thin clients that have >>> more ram, dedicated Gig-E to the new NIC on the server and better video >>> cards, for the kids whose parents are on the school board? ?That's not >>> a bad idea. ?I do have a spare PC that's at least five years >>> newer than my TC's. ?I should try one. >> >> I was thinking, from the discussion, that your bottleneck is probably >> ethernet speed, video card or RAM on the client. ?By "higher powered" I just >> meant improved in those areas. ?Might not even need full-on new systems, >> aside from the local apps angle. > > And to approach the problem from a slightly different angle, you might > try using freenx on the server with the NX client on the client as a > local app (maybe test with a windows laptop first to see if it looks > promising). ? That is very similar to normal thin-client operation in > that the desktop session is running on the server but the > proxy/caching/compression layers might cover up your bottleneck. > Having freenx installed shouldn't interfere with normal ltsp use and > it also permits easy access without rebooting from windows/mac > clients. > > -- > ? Les Mikesell > ? ? lesmikesell at gmail.com > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > From monteslu at cox.net Wed Sep 21 18:31:10 2011 From: monteslu at cox.net (Luis Montes) Date: Wed, 21 Sep 2011 11:31:10 -0700 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E791025.30001@siddall.name> References: <4E78E6F0.9090207@jomegat.com> <4E791025.30001@siddall.name> Message-ID: <4E7A2D6E.1000005@cox.net> I have a vague recollection of firefox using more client ram than it should have a few years back that cause me grief. Definitely give Chrome a shot and see if performance gets better. Jeff Siddall wrote: > On 09/20/2011 03:18 PM, Jomegat wrote: >> When he loads the web page - with no flash in it that I can find - the >> network maxes out at 12MiB/sec, and it takes an eternity for the page to >> load. He reports that sometimes this particular page will cause the TC >> to crash (black screen with text followed by login screen), but he was >> not able to reproduce that while I was there. I suspect there is some >> user error involved in the crash scenario, but it will be difficult to >> prove. > > Unfortunately there are some cases where LTSP just doesn't work well. > Sounds like you hit one of those where javascript is doing something > silly with the display and sending raw display updates across the LAN > is killing things. > > You can probably get around the problem of that one page by running > the browser as a localapp, but of course then there is no flash for > that user. > > That being said, there are a few things you can do: > > 1. Try a different browser (ex: chrome). It may work better and it > would allow you to run a dual browser (ex: firefox as a localapp and > chrome as a native). If the users will put up with it tell them to > run flash sites in Chrome and all others in Firefox. Ugly for sure > but may work for the few exception cases you are running into. > > 2. Fire up one of those clients as a Windoze machine and load a flash > page to show that Windows on your clients is a non-option. > > 3. Try installing more RAM in one client and see if it works any > better on the problem page. That should confirm/deny the running out > of RAM question. Also, that will allow you to try installing flash in > the chroot and see if the performance is acceptable. Maybe you can > add some RAM to your clients and successfully run them all with > localapp browsers. > > 4. If they are going to upgrade the PCs anyway, try to convince them > to order them diskless initially and run either LTSP with localapps or > something like DRBL. If nothing else they save admin time/effort vs. > any Windows option plus the cost of all the hard drives and Windoze > licenses in the clients. They can still add drives and Windoze after > if it doesn't work. > > Good luck! > > Jeff > > _______________________________________________ > K12OSN mailing list > K12OSN at redhat.com > https://www.redhat.com/mailman/listinfo/k12osn > For more info see > From rowens at ptd.net Wed Sep 21 23:21:59 2011 From: rowens at ptd.net (Rob Owens) Date: Wed, 21 Sep 2011 19:21:59 -0400 Subject: [K12OSN] Poor performance woes In-Reply-To: <4E791025.30001@siddall.name> References: <4E78E6F0.9090207@jomegat.com> <4E791025.30001@siddall.name> Message-ID: <20110921232159.GA26533@aurora.owens.net> On Tue, Sep 20, 2011 at 06:13:57PM -0400, Jeff Siddall wrote: > > 2. Fire up one of those clients as a Windoze machine and load a > flash page to show that Windows on your clients is a non-option. > Windows, even on an old machine, may actually fix the problem (if the problem is the network being saturated by video updates). Of course, Windows is not the cure -- any standalone machine is a possible fix. I don't fully understand the cause of your problem, but if your network connection is saturated then two possible fixes are 1) more bandwidth or 2) don't use the network to transmit video -Rob From robark at gmail.com Fri Sep 23 20:58:17 2011 From: robark at gmail.com (Robert Arkiletian) Date: Fri, 23 Sep 2011 13:58:17 -0700 Subject: [K12OSN] OT: Open Source SIS initiative launched in BC Canada Message-ID: https://groups.google.com/group/bcfosss/browse_thread/thread/fd42d61d8ba74e6d The pdf attached to the first post has all the details. -- Robert Arkiletian Eric Hamber Secondary, Vancouver, Canada