From hardik at statementofpurpose.com Wed Dec 1 17:41:00 2010 From: hardik at statementofpurpose.com (Hardik Modi) Date: Wed, 1 Dec 2010 12:41:00 -0500 Subject: Drivers, initrd and kickstart Message-ID: I use kickstart to do mostly-unattended imaging on systems that aren't network-connected. This generally works great, but I'm a little puzzled about some driver issues I'm facing right now. The base disk is CentOS 5.3 and the systems are generally SuperMicro and IBM xSeries servers. I've just had to add a new megaraid_sas driver to work with a new LSI controller. To do this, I took apart the driver update disk provided by IBM and in the stock initrd, I replaced the .ko and merged in the pci.ids and modules* files. The net effect of all this is that the controller is recognized on the IBM system and boots just fine. I do an rpm install with the new driver during the post section and the system operates correctly there after. The catch is that I can't boot up a SuperMicro system with the 3ware RAID controller with this disk. I don't see the 3w-9xxx and ata-piix drivers getting loaded and then it gets stuck at the 'CentOS disk not found' screen. Absent my changes, the system boots just fine. I've spent time looking through the files I've modified and am certain that only the LSI information has been added and it's not in conflict with the 3ware entries, in terms of IDs and dependencies. modules.cgz and initrd.img are both generated files, so I don't see a space issue, the way one might have in RHEL/CentOS 4, where you had to dd a fixed block count to generate the initrd. I know the kickstart is getting loaded, so I guess I could force the driver to be loaded. I've just realised that I could use the driverdisk option in my kickstart file, but I don't have network and I can't see a way to force it to pick up the file from the boot disk. I don't want to use separate media here, so dd from isolinux isn't really an option. Any thoughts on a better approach or ideas on what I'm doing wrong? Regards, Hardik. -------------- next part -------------- An HTML attachment was scrubbed... URL: From hardik at statementofpurpose.com Wed Dec 1 23:52:20 2010 From: hardik at statementofpurpose.com (Hardik Modi) Date: Wed, 1 Dec 2010 18:52:20 -0500 Subject: Drivers, initrd and kickstart In-Reply-To: References: Message-ID: Here's a bit of an update. I think I got it figured out over the course of the day and after much trial and error. The driver update disk came with a pcitable file and stock initrd has a pci.ids file, which I had merged (not a literal cat, but picked out what appeared to be relevant data). I've now realized that merging modules.alias is sufficient to have controllers on both kinds of systems recognized. The changes I made to pci.ids somehow were okay for the LSI controller but led to the 3ware one not being recognized. My searches on the Internet don't reveal what I ought to have done, but I'm back in business, so I'm hoping this proves useful to others in the future. If somebody on the list can shed more light on the subject, that'd be much appreciated. Hardik. On Wed, Dec 1, 2010 at 12:41 PM, Hardik Modi wrote: > I use kickstart to do mostly-unattended imaging on systems that aren't > network-connected. This generally works great, but I'm a little puzzled > about some driver issues I'm facing right now. > > The base disk is CentOS 5.3 and the systems are generally SuperMicro and > IBM xSeries servers. I've just had to add a new megaraid_sas driver to work > with a new LSI controller. To do this, I took apart the driver update disk > provided by IBM and in the stock initrd, I replaced the .ko and merged in > the pci.ids and modules* files. The net effect of all this is that the > controller is recognized on the IBM system and boots just fine. I do an rpm > install with the new driver during the post section and the system operates > correctly there after. > > The catch is that I can't boot up a SuperMicro system with the 3ware RAID > controller with this disk. I don't see the 3w-9xxx and ata-piix drivers > getting loaded and then it gets stuck at the 'CentOS disk not found' screen. > Absent my changes, the system boots just fine. > > I've spent time looking through the files I've modified and am certain that > only the LSI information has been added and it's not in conflict with the > 3ware entries, in terms of IDs and dependencies. modules.cgz and initrd.img > are both generated files, so I don't see a space issue, the way one might > have in RHEL/CentOS 4, where you had to dd a fixed block count to generate > the initrd. > > I know the kickstart is getting loaded, so I guess I could force the driver > to be loaded. I've just realised that I could use the driverdisk option in > my kickstart file, but I don't have network and I can't see a way to force > it to pick up the file from the boot disk. I don't want to use separate > media here, so dd from isolinux isn't really an option. > > Any thoughts on a better approach or ideas on what I'm doing wrong? > > Regards, > Hardik. > -------------- next part -------------- An HTML attachment was scrubbed... URL: From bakers at canbytel.com Thu Dec 2 16:52:53 2010 From: bakers at canbytel.com (Baker, Scott) Date: Thu, 2 Dec 2010 08:52:53 -0800 Subject: Controller partition order? Message-ID: I'm building a bunch of virtual machines and I want to make sure that the swap partition comes first on the disk. Sometimes we need to expand storage on VMs and this is much harder if the swap partition is after the / partition. Is there a way with the partition command to control where on the disk a partition goes? -------------- next part -------------- An HTML attachment was scrubbed... URL: From CallahanT at tessco.com Thu Dec 2 17:23:51 2010 From: CallahanT at tessco.com (Callahan, Tom) Date: Thu, 02 Dec 2010 12:23:51 -0500 Subject: Controller partition order? In-Reply-To: Message-ID: You could try ?--asprimary? in the partition command. On 12/2/10 11:52 AM, "Baker, Scott" wrote: > I'm building a bunch of virtual machines and I want to make sure that the swap > partition comes first on the disk. Sometimes we need to expand storage on VMs > and this is much harder if the swap partition is after the / partition. Is > there a way with the partition command to control where on the disk a > partition goes? > > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list -- Tom Callahan Technology, Development and Services (TDS) 410-229-1361 Tel 410-229-1512 Fax CallahanT at TESSCO.com Visit TESSCO.com TESSCO Your Total Source? for Everything Wireless Network Infrastructure Equipment | Mobile Devices & Accessories | Installation, Test, Equipment & Supplies -------------- next part -------------- An HTML attachment was scrubbed... URL: From bakers at canbytel.com Thu Dec 2 17:39:24 2010 From: bakers at canbytel.com (Scott Baker) Date: Thu, 02 Dec 2010 09:39:24 -0800 Subject: Controller partition order? In-Reply-To: References: Message-ID: <4CF7D9CC.7090800@canbytel.com> On 12/02/2010 09:23 AM, Callahan, Tom wrote: > You could try ?--asprimary? in the partition command. I only use two partitions, so I modified it as follows: # 1 gig of swap, the rest is / as ext4 part swap --size=1024 --asprimary part / --fstype=ext4 --grow --size=500 But it still creates the swap partition as /dev/vda2 and the primary / partition as /dev/vda1. Here is my full ks file if it helps http://www.perturb.org/ks/fedora.ks -- Scott Baker - Canby Telcom System Administrator - RHCE - 503.266.8253 From jared.rodriguez at rackspace.com Thu Dec 2 18:08:51 2010 From: jared.rodriguez at rackspace.com (Jared Rodriguez) Date: Thu, 2 Dec 2010 12:08:51 -0600 Subject: Controller partition order? In-Reply-To: References: Message-ID: <6580_1291313307_oB2I8MI4010347_4CF7E0B3.8070200@rackspace.com> >> I'm building a bunch of virtual machines and I want to make sure that the swap >> partition comes first on the disk. Sometimes we need to expand storage on VMs >> and this is much harder if the swap partition is after the / partition. Is >> there a way with the partition command to control where on the disk a >> partition goes? We use something like this: zerombr yes clearpart --all --initlabel part /boot --fstype "ext3" --size=100 --asprimary part /tmp --fstype "ext3" --size=2048 --asprimary part / --fstype "ext3" --size=1 --grow %include /tmp/swap.cfg swap is dynamically generated during the bootstrap process. The configuration creates the following partitions: /dev/sda1 /boot /dev/sda2 /tmp /dev/sda3 swap /dev/sda4 extended /dev/sda5 / I am not aware of any other way to make anaconda honor the position of the swap partition. jrod Confidentiality Notice: This e-mail message (including any attached or embedded documents) is intended for the exclusive and confidential use of the individual or entity to which this message is addressed, and unless otherwise expressly indicated, is confidential and privileged information of Rackspace. Any dissemination, distribution or copying of the enclosed material is prohibited. If you receive this transmission in error, please notify us immediately by e-mail at abuse at rackspace.com, and delete the original message. Your cooperation is appreciated. From flaks at bnl.gov Thu Dec 2 18:34:56 2010 From: flaks at bnl.gov (Leonid Flaks) Date: Thu, 02 Dec 2010 13:34:56 -0500 Subject: Controller partition order? In-Reply-To: <6580_1291313307_oB2I8MI4010347_4CF7E0B3.8070200@rackspace.com> References: <6580_1291313307_oB2I8MI4010347_4CF7E0B3.8070200@rackspace.com> Message-ID: <4CF7E6D0.50306@bnl.gov> On 12/02/2010 01:08 PM, Jared Rodriguez wrote: >>> I'm building a bunch of virtual machines and I want to make sure >>> that the swap >>> partition comes first on the disk. Sometimes we need to expand >>> storage on VMs >>> and this is much harder if the swap partition is after the / >>> partition. Is >>> there a way with the partition command to control where on the disk a >>> partition goes? > > We use something like this: > > zerombr yes > clearpart --all --initlabel > part /boot --fstype "ext3" --size=100 --asprimary > part /tmp --fstype "ext3" --size=2048 --asprimary > part / --fstype "ext3" --size=1 --grow > %include /tmp/swap.cfg > > swap is dynamically generated during the bootstrap process. The > configuration creates the following partitions: > > /dev/sda1 /boot > /dev/sda2 /tmp > /dev/sda3 swap > /dev/sda4 extended > /dev/sda5 / > > I am not aware of any other way to make anaconda honor the position of > the swap partition. > > jrod We use somewhat similar solution: part --onpart=sda1 swap part --onpart=sda2 / --fstype ext4 For that to work in %pre section we run fdisk, which we feed with file that has all fdisk commands written one per line. %pre /usr/bin/wget -nv http://myserver/path/to/file/fdisk.input -O /tmp/fdisk.input /usr/sbin/fdisk /dev/sda< /tmp/fdisk.input Leon From jcm at redhat.com Thu Dec 2 19:25:48 2010 From: jcm at redhat.com (Jon Masters) Date: Thu, 02 Dec 2010 11:25:48 -0800 Subject: Drivers, initrd and kickstart In-Reply-To: References: Message-ID: <1291317951.2227.12.camel@monticello.bos.jonmasters.org> On Wed, 2010-12-01 at 18:52 -0500, Hardik Modi wrote: > If somebody on the list can shed more light on the subject, that'd be much > appreciated. If you are running on RHEL, feel free to drop me a line off-list. Jon. From no-reply at dropboxmail.com Fri Dec 3 06:10:05 2010 From: no-reply at dropboxmail.com (Dropbox) Date: Fri, 03 Dec 2010 06:10:05 +0000 Subject: Ej Magpayo invited you to Dropbox Message-ID: <20101203061005.7126D37EC9E@mailman-2.dropboxmail.com> Ej Magpayo wants you to use Dropbox to sync and share files online and across computers. Get started here: http://www.dropbox.com/link/20.Pcmmi_5ntW/NjUwMzE3NjEzNw?src=referrals_bulk0 - The Dropbox Team ____________________________________________________ To stop receiving invites from Dropbox, please go to http://www.dropbox.com/bl/9b2ab5078ffb/kickstart-list%40redhat.com -------------- next part -------------- An HTML attachment was scrubbed... URL: From ks-list at puzzled.xs4all.nl Fri Dec 3 16:59:02 2010 From: ks-list at puzzled.xs4all.nl (Patrick Lists) Date: Fri, 03 Dec 2010 17:59:02 +0100 Subject: Use port range for multiple ports in firewall possible? Message-ID: <4CF921D6.6060404@puzzled.xs4all.nl> Hi, Reading through the Kickstart docs I wonder if it is possible to do the following. Udp port 5060 is the SIP signaling port and the 10000-10050 range are the rtp ports: firewall --enabled --ssh --port=5060:udp --port=10000-10050:udp So specify a port range for multiple consecutive ports instead of a comma separated list. In this case a comma separated list would not be very practical. If it is not possible to do this how do you handle this? In %post? Thanks and regards, Patrick From andy.speagle at wichita.edu Sat Dec 11 17:18:00 2010 From: andy.speagle at wichita.edu (Speagle, Andy) Date: Sat, 11 Dec 2010 11:18:00 -0600 Subject: RHEL 6 Anaconda Error - AttributeError Message-ID: Hi Team, I'm setting up my RHEL 6 Kickstarts within my Spacewalk server and am running into this error while running the "traceback" scripts: CRITICAL: anaconda 13.21.82 exception report Traceback (most recent call first): File "/usr/lib/anaconda/instdata.py", line 172, in write services = list(self.storage.services) File "/usr/lib/anaconda/backend.py", line 316, in writeConfiguration anaconda.id.write() File "/usr/lib/anaconda/dispatch.py", line 208, in moveStep rc = stepFunc(self.anaconda) File "/usr/lib/anaconda/dispatch.py", line 126, in gotoNext self.moveStep() File "/usr/lib/anaconda/text.py", line 657, in run anaconda.dispatch.gotoNext() File "/usr/bin/anaconda", line 1115, in anaconda.intf.run(anaconda) AttributeError: 'Storage' object has no attribute 'services' Has anyone seen this? Any clues as to where I should begin? Thanks, Andy Speagle System & Storage Administrator UCATS - Wichita State University P: 316.978.3869 C: 316.617.2431 -------------- next part -------------- An HTML attachment was scrubbed... URL: From clumens at redhat.com Mon Dec 13 15:52:22 2010 From: clumens at redhat.com (Chris Lumens) Date: Mon, 13 Dec 2010 10:52:22 -0500 Subject: RHEL 6 Anaconda Error - AttributeError In-Reply-To: References: Message-ID: <20101213155222.GL30483@exeter.usersys.redhat.com> > CRITICAL: anaconda 13.21.82 exception report > Traceback (most recent call first): > File "/usr/lib/anaconda/instdata.py", line 172, in write > services = list(self.storage.services) > File "/usr/lib/anaconda/backend.py", line 316, in writeConfiguration > anaconda.id.write() > File "/usr/lib/anaconda/dispatch.py", line 208, in moveStep > rc = stepFunc(self.anaconda) > File "/usr/lib/anaconda/dispatch.py", line 126, in gotoNext > self.moveStep() > File "/usr/lib/anaconda/text.py", line 657, in run > anaconda.dispatch.gotoNext() > File "/usr/bin/anaconda", line 1115, in > anaconda.intf.run(anaconda) > AttributeError: 'Storage' object has no attribute 'services' > > Has anyone seen this? Any clues as to where I should begin? Can you paste the entire /tmp/anaconda-tb-* file somewhere? This doesn't look possible. - Chris From jnaikwadi at asterdata.com Tue Dec 14 18:53:57 2010 From: jnaikwadi at asterdata.com (Jay Naikwadi) Date: Tue, 14 Dec 2010 10:53:57 -0800 Subject: RAID chunk size configuration from kickstart Message-ID: <4D07BD45.9080204@asterdata.com> Hi , I am trying to deploy software raid setup using kickstart and was wondering if I can specify chunk-size as parameter in kickstart file.I am using RHEL 6. I didn't see any documentation for this parameter in kickstart. The default installation seems to be using 512K as chunk size whereas I would like to set it to 8MB . This is new device configuration and there is nothing configured on system before this. If kickstart does not allow setting this parameter , Is there another way in %pre , %post sections where we can change it ? How else this parameter could be set for '/' , '/boot/' partitions on clean install ? Thanks and regards, Jay -------------- next part -------------- An HTML attachment was scrubbed... URL: From andy.speagle at wichita.edu Tue Dec 14 21:12:36 2010 From: andy.speagle at wichita.edu (Speagle, Andy) Date: Tue, 14 Dec 2010 15:12:36 -0600 Subject: RHEL 6 Anaconda Error - AttributeError In-Reply-To: <20101213155222.GL30483@exeter.usersys.redhat.com> References: <20101213155222.GL30483@exeter.usersys.redhat.com> Message-ID: > > CRITICAL: anaconda 13.21.82 exception report Traceback (most recent > > call first): > > File "/usr/lib/anaconda/instdata.py", line 172, in write > > services = list(self.storage.services) > > File "/usr/lib/anaconda/backend.py", line 316, in writeConfiguration > > anaconda.id.write() > > File "/usr/lib/anaconda/dispatch.py", line 208, in moveStep > > rc = stepFunc(self.anaconda) > > File "/usr/lib/anaconda/dispatch.py", line 126, in gotoNext > > self.moveStep() > > File "/usr/lib/anaconda/text.py", line 657, in run > > anaconda.dispatch.gotoNext() > > File "/usr/bin/anaconda", line 1115, in > > anaconda.intf.run(anaconda) > > AttributeError: 'Storage' object has no attribute 'services' > > > > Has anyone seen this? Any clues as to where I should begin? > > Can you paste the entire /tmp/anaconda-tb-* file somewhere? This doesn't > look possible. Here's a link to the anaconda-tb-* ... http://www.aqr.aero/files/anaconda-tb-tTx5Tj Thanks, Andy Speagle From thomas.espe at gmail.com Wed Dec 15 12:16:15 2010 From: thomas.espe at gmail.com (Thomas Hemmingby Espe) Date: Wed, 15 Dec 2010 13:16:15 +0100 Subject: Problem with fdisk partition aligment (for VMWare) and kickstart patitioning Message-ID: (this has also been posted to rhn-satellite-users, sorry for the cross-posting). Hello. We are trying to setup automated partition aligment during kickstart install. The kickstart is generated from a Satellite 5.4 installation. In the %pre-section of the kickstart, I have the following script: fdisk /dev/sda < References: Message-ID: <4D08C208.6040804@control.lth.se> On 2010-12-15 13.16, Thomas Hemmingby Espe wrote: > (this has also been posted to rhn-satellite-users, sorry for the cross-posting). > > Hello. > > We are trying to setup automated partition aligment during kickstart > install. The kickstart is generated > from a Satellite 5.4 installation. > > In the %pre-section of the kickstart, I have the following script: > > fdisk /dev/sda < d > 1 > d > 2 > d > 3 > d > 4 > n > p > 1 > 1 > +200m > t > fb > x > b > 1 > 128 > w > EOF > > which deletes previously created partitions, creates a 200MB partition > and aligns it to sector 128, as per > http://www.vmware.com/pdf/esx3_partition_align.pdf > Then, we create the remaning partitions: > > part /boot --fstype=ext3 --noformat --onpart=sda1 I think formatting the just created partition might help. > part pv.00 --size=1000 --grow > volgroup vg00 pv.00 > logvol /opt --vgname=vg00 --name=optvol --size=6000 --grow > logvol /var/opt --vgname=vg00 --name=varoptvol --size=1000 --grow > logvol /var --vgname=vg00 --name=varvol --size=4000 > logvol / --vgname=vg00 --name=rootvol --size=8000 > logvol swap --vgname=vg00 --name=swap --size=4000 > > During installation, after the %pre-scripts have run, the following > messages appears: > > "Unable to locate partition sda1 to use for /boot. > Press OK to reboot your system" > > On VT 2, I am able to verify that the partition does indeed exist > (using fdisk -lu) and that it starts at sector 128. > > Has anyone else been experiencing this? Does anyone see a solution to this? > > Sincerely, > > > -- Anders Blomdell Email: anders.blomdell at control.lth.se Department of Automatic Control Lund University Phone: +46 46 222 4625 P.O. Box 118 Fax: +46 46 138118 SE-221 00 Lund, Sweden From thomas.espe at gmail.com Wed Dec 15 14:39:28 2010 From: thomas.espe at gmail.com (Thomas Hemmingby Espe) Date: Wed, 15 Dec 2010 15:39:28 +0100 Subject: Problem with fdisk partition aligment (for VMWare) and kickstart patitioning In-Reply-To: <4D08C208.6040804@control.lth.se> References: <4D08C208.6040804@control.lth.se> Message-ID: 2010/12/15 Anders Blomdell : >> part /boot --fstype=ext3 --noformat --onpart=sda1 > I think formatting the just created partition might help. Thanks for the answer, but it did not help either. -- Thomas H. Espe From hbrown at divms.uiowa.edu Wed Dec 15 15:19:46 2010 From: hbrown at divms.uiowa.edu (Hugh Brown) Date: Wed, 15 Dec 2010 09:19:46 -0600 Subject: Problem with fdisk partition aligment (for VMWare) and kickstart patitioning In-Reply-To: References: Message-ID: <4D08DC92.9070704@divms.uiowa.edu> Thomas Hemmingby Espe wrote: > (this has also been posted to rhn-satellite-users, sorry for the cross-posting). > > Hello. > > We are trying to setup automated partition aligment during kickstart > install. The kickstart is generated > from a Satellite 5.4 installation. > > In the %pre-section of the kickstart, I have the following script: > > fdisk /dev/sda < d > 1 > d > 2 > d > 3 > d > 4 > n > p > 1 > 1 > +200m > t > fb > x > b > 1 > 128 > w > EOF > > which deletes previously created partitions, creates a 200MB partition > and aligns it to sector 128, as per > http://www.vmware.com/pdf/esx3_partition_align.pdf > Then, we create the remaning partitions: > > part /boot --fstype=ext3 --noformat --onpart=sda1 > part pv.00 --size=1000 --grow > volgroup vg00 pv.00 > logvol /opt --vgname=vg00 --name=optvol --size=6000 --grow > logvol /var/opt --vgname=vg00 --name=varoptvol --size=1000 --grow > logvol /var --vgname=vg00 --name=varvol --size=4000 > logvol / --vgname=vg00 --name=rootvol --size=8000 > logvol swap --vgname=vg00 --name=swap --size=4000 > > During installation, after the %pre-scripts have run, the following > messages appears: > > "Unable to locate partition sda1 to use for /boot. > Press OK to reboot your system" > > On VT 2, I am able to verify that the partition does indeed exist > (using fdisk -lu) and that it starts at sector 128. > > Has anyone else been experiencing this? Does anyone see a solution to this? > > Sincerely, > > > Instead of doing the "d 1 d 2 ..." I just zero out the mbr with dd dd if=/dev/zero of=/dev/sda bs=512 count=1 Then, I do a similar fdisk < References: <4D07BD45.9080204@asterdata.com> Message-ID: <4D0912A4.4020609@asterdata.com> Jay Naikwadi wrote: > Hi , > > I am trying to deploy software raid setup using kickstart and was > wondering if I can specify chunk-size as parameter in kickstart file.I > am using RHEL 6. I didn't see any documentation for this parameter in > kickstart. The default installation seems to be using 512K as chunk > size whereas I would like to set it to 8MB . This is new device > configuration and there is nothing configured on system before this. > > If kickstart does not allow setting this parameter , Is there another > way in %pre , %post sections where we can change it ? How else this > parameter could be set for '/' , '/boot/' partitions on clean > install ? > > Thanks and regards, > Jay Any suggestions about how this could be done ? -Jay -------------- next part -------------- An HTML attachment was scrubbed... URL: From thomas.espe at gmail.com Wed Dec 15 20:03:36 2010 From: thomas.espe at gmail.com (Thomas Hemmingby Espe) Date: Wed, 15 Dec 2010 21:03:36 +0100 Subject: Problem with fdisk partition aligment (for VMWare) and kickstart patitioning In-Reply-To: <4D08DC92.9070704@divms.uiowa.edu> References: <4D08DC92.9070704@divms.uiowa.edu> Message-ID: 2010/12/15 Hugh Brown : Hello, and thanks for the reply. > Instead of doing the "d 1 d 2 ..." I just zero out the mbr with dd > > dd if=/dev/zero of=/dev/sda bs=512 count=1 Yes, we changed the script today to something similar to what you suggest, but that did not make any difference. > Then, I do a similar fdisk < and create a partition for the physical volume, set it to making sure to > start/end on a boundary. > > I'd double check to make sure you don't have a clearpart directive that is > overwriting your changes. > > You could also try running partprobe after you finish writing out the change > to the partition table. Somehow, partprobe is not available. I have not manage to find out why. RedHat suggests that partprobe is part of the parted-package, and parted is available, but not partprobe. The installation halts just after the %pre-script is done (the repartitioning is the only %pre-script that is run), so it seems that anaconda is somehow not able to pick up the changes. partprobe could have remedied that, I believe, had it been available. -- Thomas H. Espe From hbrown at divms.uiowa.edu Wed Dec 15 21:36:50 2010 From: hbrown at divms.uiowa.edu (Hugh Brown) Date: Wed, 15 Dec 2010 15:36:50 -0600 Subject: Problem with fdisk partition aligment (for VMWare) and kickstart patitioning In-Reply-To: References: <4D08DC92.9070704@divms.uiowa.edu> Message-ID: <4D0934F2.5030402@divms.uiowa.edu> Thomas Hemmingby Espe wrote: > 2010/12/15 Hugh Brown : > > Hello, and thanks for the reply. > >> Instead of doing the "d 1 d 2 ..." I just zero out the mbr with dd >> >> dd if=/dev/zero of=/dev/sda bs=512 count=1 > > Yes, we changed the script today to something similar to what you suggest, > but that did not make any difference. > >> Then, I do a similar fdisk <> and create a partition for the physical volume, set it to making sure to >> start/end on a boundary. >> >> I'd double check to make sure you don't have a clearpart directive that is >> overwriting your changes. >> >> You could also try running partprobe after you finish writing out the change >> to the partition table. > > Somehow, partprobe is not available. I have not manage to find out > why. RedHat suggests that > partprobe is part of the parted-package, and parted is available, but > not partprobe. > > The installation halts just after the %pre-script is done (the > repartitioning is the only > %pre-script that is run), so it seems that anaconda is somehow not > able to pick up the changes. > partprobe could have remedied that, I believe, had it been available. > Which version of Redhat are you running? I've been doing this with VMs using RHEL5 running on ESX 4.0 with the storage device presented to vmware via NFS from a NAS box. Hugh From jamesaharrisonuk at yahoo.co.uk Wed Dec 15 23:46:02 2010 From: jamesaharrisonuk at yahoo.co.uk (jamesaharrisonuk at yahoo.co.uk) Date: Wed, 15 Dec 2010 23:46:02 +0000 Subject: Problem with fdisk partition aligment (for VMWare) and kickstartpatitioning In-Reply-To: <4D0934F2.5030402@divms.uiowa.edu> References: <4D08DC92.9070704@divms.uiowa.edu><4D0934F2.5030402@divms.uiowa.edu> Message-ID: <1940144498-1292456741-cardhu_decombobulator_blackberry.rim.net-674600265-@bda2469.bisx.prod.on.blackberry> I don't have these problems with kvm. Using rhel 5.5. Sent from my Verizon Wireless BlackBerry -----Original Message----- From: Hugh Brown Sender: kickstart-list-bounces at redhat.com Date: Wed, 15 Dec 2010 15:36:50 To: Discussion list about Kickstart Reply-To: Discussion list about Kickstart Subject: Re: Problem with fdisk partition aligment (for VMWare) and kickstart patitioning Thomas Hemmingby Espe wrote: > 2010/12/15 Hugh Brown : > > Hello, and thanks for the reply. > >> Instead of doing the "d 1 d 2 ..." I just zero out the mbr with dd >> >> dd if=/dev/zero of=/dev/sda bs=512 count=1 > > Yes, we changed the script today to something similar to what you suggest, > but that did not make any difference. > >> Then, I do a similar fdisk <> and create a partition for the physical volume, set it to making sure to >> start/end on a boundary. >> >> I'd double check to make sure you don't have a clearpart directive that is >> overwriting your changes. >> >> You could also try running partprobe after you finish writing out the change >> to the partition table. > > Somehow, partprobe is not available. I have not manage to find out > why. RedHat suggests that > partprobe is part of the parted-package, and parted is available, but > not partprobe. > > The installation halts just after the %pre-script is done (the > repartitioning is the only > %pre-script that is run), so it seems that anaconda is somehow not > able to pick up the changes. > partprobe could have remedied that, I believe, had it been available. > Which version of Redhat are you running? I've been doing this with VMs using RHEL5 running on ESX 4.0 with the storage device presented to vmware via NFS from a NAS box. Hugh _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From S.Reber at lcsys.ch Thu Dec 16 18:14:43 2010 From: S.Reber at lcsys.ch (Simon Reber) Date: Thu, 16 Dec 2010 19:14:43 +0100 Subject: RHEL 6 Kickstart with interactive part does not work Message-ID: <1209832B38DC214CB373A59426B91DFEB3460B@chbsex01.lcsys.ch> Hi all, I'm currently trying to make my RHEL 5 kickstart environment working for RHEL 6. So far everything looks OK - but when I use the special kickstart file, that takes some arguments from the user it fails. The problem is that I have to change the tty (chvt) to get the interactive stuff working - but for some reason doesn't it have the same behavior as we had in RHEL 5. It still changes the tty but after that, the script doesn't process further. I personally think that it has something to-do with the tty change (but I also can be wrong). So if somebody has a working example, I would be really grateful as I've tried almost everything without luck. My kickstart file is build as follow: # Test Kickstart # Installation specification install # url --url http://url_to_repo cdrom text key --skip lang en_US.UTF-8 keyboard sg-latin1 %include /tmp/network-include rootpw --iscrypted xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx firewall --disabled authconfig --enableshadow --enablemd5 selinux --disabled timezone --utc Europe/Zurich firstboot --disabled logging --level=debug # Partition specification %include /tmp/part-include reboot %packages --excludedocs --nobase kernel yum openssh-server openssh-clients audit man logrotate tmpwatch vixie-cron crontabs ksh ntp perl bind-utils sudo bc -sysstat which sendmail wget xinetd redhat-lsb dhclient -kexec-tools -firstboot -rhnsd -tftp-server -system-config-soundcard -squashfs-tools -device-mapper-multipath -aspell-en -aspell -rdate -dhcpv6-client -NetworkManager -rsh -sysreport -irda-utils -rdist -anacron -bluez-utils -talk -system-config-lvm -wireless-tools -setroubleshoot-server -setroubleshoot-plugins -setroubleshoot -ppp -GConf2 -dhcpv6-client -iptables-ipv6 -libselinux-python -setools -selinux-policy -libselinux-utils -chkfontpath -urw-fonts -xorg-x11-xfs -policycoreutils -selinux-policy-targeted -ypbind -yp-tools -smartmontools -pcsc-lite -trousers -oddjob -yum-updatesd -rhnsd -readahead -pcsc-lite -gpm -at -cpuspeed -conman -system-config-securitylevel-tui -tcsh -firstboot-tui -ppp -rp-pppoe -system-config-network-tui -syslinux -pcsc-lite-libs -pcmciautils -pam_smb -mkbootdisk -jwhois -ipsec-tools -ed -authconfig -crash -Deployment_Guide-en-US -hal -pm-utils -dbus -dbus-glib %pre --interpreter python import os import sys import re import os.path import time os.system('chvt 3') os.system('exec < /dev/tty3 > /dev/tty3') os.system('clear') # Defining file name and location file_network = '/tmp/network-include' file_part = '/tmp/part-include' file_lcsetup = '/tmp/LCSetup' # Default Host and DNS domain hostname = 'localhost' dns = 'localdomain' # Get current date/time localtime = time.asctime( time.localtime(time.time()) ) # Function definition def ipFormatChk(ip_addr): pattern = r"\b(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0 -9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9] |[01]?[0-9][0-9]?)\b" if re.match(pattern, ip_addr): return True else: return False def ProtoChk(proto): if proto == 'dhcp' or proto == 'static': return True else: return False def GetSerial(dmi): if re.search('Serial Number', dmi): return True else: return False def CheckLayout(val): if val == 'cmp' or val == 'db' or val == 'opt': return True else: return False def GetPartition(val1,val2): # split disk information toto device and # size part_size_mb = 0 dev, size_mb = val2.split(";") dev = re.sub('\/dev\/','',dev) array = list() array.append('bootloader --location=mbr --driveorder=%s --append=\"audit=1 crashkernel=auto\"' % dev) array.append('clearpart --linux --drives=%s' % dev) array.append('part /boot --fstype ext4 --size=150 --ondisk=%s' % dev) # Enter swap size swap = raw_input('Swap size in MB: ') while not re.match("^[0-9]+$", swap): swap = raw_input('Invalide argument, please try again in MB [numeric]: ') # Write swap into integer for calcultion swap_int = int(swap) array.append('part swap --size=%s --ondisk=%s' % (swap, dev)) array.append('part pv.01 --grow --size=1 --ondisk=%s' % dev) array.append('volgroup vgroot pv.01') if val1 == 'cmp': array.append('logvol / --vgname=vgroot --size=3072 --name=lvroot --fstype=ext4') array.append('logvol /var --vgname=vgroot --size=2096 --name=lvvar --fstype=ext4') array.append('logvol /tmp --vgname=vgroot --size=1024 --name=lvtmp --fstype=ext4') array.append('logvol /home --vgname=vgroot --size=1024 --name=lvhome --fstype=ext4') array.append('logvol /opt/spare --vgname=vgroot --size=1 --grow --name=lvspare --fstype=ext4') part_size_mb = swap_int + 7264 elif val1 == 'db': array.append('logvol / --vgname=vgroot --size=3072 --name=lvroot --fstype=ext4') array.append('logvol /var --vgname=vgroot --size=2096 --name=lvvar --fstype=ext4') array.append('logvol /tmp --vgname=vgroot --size=1024 --name=lvtmp --fstype=ext4') array.append('logvol /home --vgname=vgroot --size=1024 --name=lvhome --fstype=ext4') array.append('logvol /opt/oracle --vgname=vgroot --size=1 --grow --name=lvoracle --fstype=ext4') part_size_mb = swap_int + 7264 elif val1 == 'opt': array.append('logvol / --vgname=vgroot --size=3072 --name=lvroot --fstype=ext4') array.append('logvol /var --vgname=vgroot --size=2096 --name=lvvar --fstype=ext4') array.append('logvol /opt --vgname=vgroot --size=3072 --name=lvopt --fstype=ext4') array.append('logvol /tmp --vgname=vgroot --size=1024 --name=lvtmp --fstype=ext4') array.append('logvol /home --vgname=vgroot --size=1024 --name=lvhome --fstype=ext4') array.append('logvol /opt/spare --vgname=vgroot --size=1 --grow --name=lvspare --fstype=ext4') part_size_mb = swap_int + 10336 part_size_mb = int(part_size_mb) size_mb = int(size_mb) if part_size_mb > size_mb: return (array,False) else: return (array,True) # Main Part print ('Welcome to LCSetup\n') # Trying to get serial number using # dmidecode demidecode = os.popen("/usr/sbin/dmidecode") sysinfo = demidecode.readlines() err = demidecode.close() # Get first Serial Number information # since this usually is the serial number of # the computer c = 1 b = '' for line in sysinfo: i = GetSerial(line) if i == True and c == 1: a, b = line.split(":") b = b.strip() c+=1 if b: serial = raw_input('Serial number [%s]: ' % b) else: serial = raw_input('Serial number: ') if not serial and b: serial = b else: while not serial: serial = raw_input('Invalide serial number, try again: ') # Ask for the hostname hostname = raw_input('Hostname of Machine [%s]: ' % hostname) if hostname == '': hostname = 'localhost' # Write information into LCSetup file lcs = open(file_lcsetup, 'w') lcs.write('### LCSetup kickstart installation (%s)\n' % localtime) lcs.write('###\n') lcs.write('SERIAL: %s\n' % serial) lcs.write('HOSTNAME: %s\n' % hostname) lcs.close # Network configuration ifconfig = os.popen("/usr/sbin/ifconfig") nic = ifconfig.readlines() err = ifconfig.close() eth = '' for n in nic: n = n.strip() if re.search('^eth', n): devif, t = n.split("Link") devif = devif.strip() net_if = raw_input('Configure interface %s [- to skip]: ' % devif) if net_if != '-': eth = devif break else: continue if net_if == '-' or net_if == '': net_if = 'localhost' if eth != '': proto = raw_input('DHCP or Static [dhcp|static]: ') while not ProtoChk(proto): proto = raw_input('Please enter [dhcp] or [static]: ') if proto == 'dhcp': dns = raw_input('DNS domain [%s]: ' % dns) if dns == '': dns = 'localdomain' elif proto == 'static': ip_addr = raw_input('IP address: ') while not ipFormatChk(ip_addr): ip_addr = raw_input('Invalide IP address, try again: ') subnet_addr = raw_input('Subnet address: ') while not ipFormatChk(subnet_addr): subnet_addr = raw_input('Invalide subnet address, try again: ') default_gw = raw_input('Default gateway: ') while not ipFormatChk(default_gw): default_gw = raw_input('Invalide default gateway, try again: ') dns = raw_input('DNS domain [%s]: ' % dns) if dns == '': dns = 'localdomain' else: print "Unknown error...\n" sys.exit(1) # Write information to file if proto == 'dhcp': file = open(file_network, 'w') file.write('network --device %s --bootproto %s\n' % (eth, proto)) file.close elif proto == 'static': file = open(file_network, 'w') file.write('network --device=%s --bootproto=%s --ip=%s --netmask=%s --gateway=%s --nameserver=xxxxxxxxx --hostname=%s' % (eth, proto, ip_addr, subnet_addr, default_gw, hostname)) file.close else: print "Unknown error..." sys.exit(1) # Write information into LCSetup file lcs = open(file_lcsetup, 'a') lcs.write('DNS_DOMAIN: %s\n' % dns) lcs.write('NETWORKING=yes\n') lcs.write('NET_DEVICE_%s: %s\n' % (eth, eth)) lcs.write('NET_PROTOCOL_%s: %s\n' % (eth, proto)) if proto == 'static': lcs.write('NET_IP_ADDR_%s: %s\n' % (eth, ip_addr)) lcs.write('NET_SUB_ADDR_%s: %s\n' % (eth, subnet_addr)) lcs.write('NET_DEFAULT_GW: %s\n' % default_gw) lcs.close else: lcs = open(file_lcsetup, 'a') lcs.write('NETWORKING=no') lcs.close file = open(file_network, 'w') file.close # Get boot parameter and select the apropriate # partition layout disk_size = 0 sfdisk = os.popen("/usr/sbin/sfdisk -s") disk = sfdisk.readlines() err = sfdisk.close() c = 0 counter = 1 for count in disk: if re.search('\/dev\/',count): c += 1 disk_array = list() for line in disk: if re.search('\/dev\/',line): if c <= 1: dev, size = line.split(':') dev = dev.strip() size = size.strip() size = int(size) size_mb = size / 1024 size_mb = str(size_mb) tmp = dev + ";" + size_mb # print ("%s and %s MB" % (dev, size_mb)) disk_size = tmp else: dev, size = line.split(':') dev = dev.strip() size = size.strip() size = int(size) size_mb = size / 1024 if counter == 1: print "The following disks are available for usage:\n" print "\t[%s] %s with size %s MB" % (counter, dev, size_mb) size_mb = str(size_mb) tmp = dev + ";" + size_mb disk_array.append(tmp) counter += 1 if c >= 2: disk_selection = raw_input('\nEnter ID of disk to take: ') disk_selection = int(disk_selection) disk_selection -= 1 disk_size = disk_array[disk_selection] # Special workaround is required to refuse # not size-able partition layout hd_size = False part_layout = raw_input('\nEnter partition layout [cmp|db|opt]: ') while not CheckLayout(part_layout): part_layout = raw_input('Invalide partition layout, try again [cmp|db|opt]: ') part, hd_size = GetPartition(part_layout, disk_size) while not hd_size: part_layout = raw_input('Size of partition layout does not fit, choose an other [cmp|db|opt]: ') while not CheckLayout(part_layout): part_layout = raw_input('Invalide partition layout, try again [cmp|db|opt]: ') part, hd_size = GetPartition(part_layout, disk_size) # Write partition table to file file = open(file_part, 'w') for l in part: l = l.strip() file.write("%s\n" % l) file.close # Write partition information to LCSetup # information disk_dev, disk_dev_sz = disk_size.split(";") lcs = open(file_lcsetup, 'a') lcs.write('PART_LAYOUT: %s\n' % part_layout) lcs.write('ROOT_DISK: %s\n' % disk_dev) lcs.close os.system('chvt 1') os.system('exec < /dev/tty1 > /dev/tty1') %post --nochroot chvt 2 exec < /dev/tty2 > /dev/tty2 mkdir -p /mnt/sysimage/usr/local/host/etc chmod -R 750 /mnt/sysimage/usr/local/host cp /tmp/LCSetup /mnt/sysimage/usr/local/host/etc # chroot /mnt/sysimage /usr/local/host/bin/LCSetup # rm -rf /mnt/sysimage/usr/local/host/tools/* chvt 1 exec < /dev/tty1 > /dev/tty1 %end Thanks and all the best, Simon Reber From cjk at techma.com Sat Dec 18 07:04:28 2010 From: cjk at techma.com (Kovacs, Corey J.) Date: Sat, 18 Dec 2010 07:04:28 +0000 Subject: RAID chunk size configuration from kickstart In-Reply-To: <4D0912A4.4020609@asterdata.com> References: <4D07BD45.9080204@asterdata.com>,<4D0912A4.4020609@asterdata.com> Message-ID: As far as I know, not for the root fs which "normally" shouldn't be a problem since setting the chunk size is a performance tuning task and you typically don't do high load tasks against your root fs (Well I don't anyway). So, I'd suggest simply building a small root fs using the normal kick start methods, then elsewhere in your post section, manually create the raid device and file-system where you need to set the chunk size. Also, this is sort of predicated by all of your nodes are exactly the same and have exactly the same (or very similar) workload. -C ________________________________ From: kickstart-list-bounces at redhat.com [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi [jnaikwadi at asterdata.com] Sent: Wednesday, December 15, 2010 2:10 PM To: kickstart-list at redhat.com Subject: Re: RAID chunk size configuration from kickstart Jay Naikwadi wrote: Hi , I am trying to deploy software raid setup using kickstart and was wondering if I can specify chunk-size as parameter in kickstart file.I am using RHEL 6. I didn't see any documentation for this parameter in kickstart. The default installation seems to be using 512K as chunk size whereas I would like to set it to 8MB . This is new device configuration and there is nothing configured on system before this. If kickstart does not allow setting this parameter , Is there another way in %pre , %post sections where we can change it ? How else this parameter could be set for '/' , '/boot/' partitions on clean install ? Thanks and regards, Jay Any suggestions about how this could be done ? -Jay -------------- next part -------------- An HTML attachment was scrubbed... URL: From jared.rodriguez at rackspace.com Mon Dec 20 23:20:29 2010 From: jared.rodriguez at rackspace.com (Jared Rodriguez) Date: Mon, 20 Dec 2010 17:20:29 -0600 Subject: Anaconda stage 1 and ssl Message-ID: <4325_1292887138_oBKNIqQP008913_4D0FE4BD.1040604@rackspace.com> I am attempting to host my ks.cfg file on an SSL enabled web server. Unfortunately, the installer does not seem to have http-ssl support. Am I missing an option? Or is this the sad truth? The user agent is reported as: "anaconda/11.1.2.113" Jared Confidentiality Notice: This e-mail message (including any attached or embedded documents) is intended for the exclusive and confidential use of the individual or entity to which this message is addressed, and unless otherwise expressly indicated, is confidential and privileged information of Rackspace. Any dissemination, distribution or copying of the enclosed material is prohibited. If you receive this transmission in error, please notify us immediately by e-mail at abuse at rackspace.com, and delete the original message. Your cooperation is appreciated. From ks-list at puzzled.xs4all.nl Tue Dec 21 12:18:37 2010 From: ks-list at puzzled.xs4all.nl (Patrick Lists) Date: Tue, 21 Dec 2010 13:18:37 +0100 Subject: Detecting amount of memory? Message-ID: <4D109B1D.40500@puzzled.xs4all.nl> Hi, I trying to create a kickstart script on CentOS 5.5 that installs an RPM depending on the amount of memory in the box. For example if the box has >= 6GB memory then install RPM2 else install RPM1. With some Googling I now have the following snippet: %pre memKB=$(free|grep Mem|awk '{print$2}') memGB=$(echo "($memKB/1024)/1024"|bc) BIGMEM="" if [ "$memGB" -ge 6 ]; then BIGMEM="6GB" fi %post yum install foobar$BIGMEM-1.0-1.rpm The problem is that on a box with 6GB memory this snippet says it does not have 6GB mem: memKB = 6123716 KB memGB = 5 GB Anyone know what I am doing wrong? Thanks! Patrick From jamesaharrisonuk at yahoo.co.uk Tue Dec 21 12:39:16 2010 From: jamesaharrisonuk at yahoo.co.uk (jamesaharrisonuk at yahoo.co.uk) Date: Tue, 21 Dec 2010 12:39:16 +0000 Subject: Detecting amount of memory? In-Reply-To: <4D109B1D.40500@puzzled.xs4all.nl> References: <4D109B1D.40500@puzzled.xs4all.nl> Message-ID: <591391867-1292935156-cardhu_decombobulator_blackberry.rim.net-1942212958-@bda2469.bisx.prod.on.blackberry> The kernel reports the amount of memory in /proc/meninfo . It shows the memory information in kilobytes. James Sent from my Verizon Wireless BlackBerry -----Original Message----- From: Patrick Lists Sender: kickstart-list-bounces at redhat.com Date: Tue, 21 Dec 2010 13:18:37 To: Discussion list about Kickstart Reply-To: Discussion list about Kickstart Subject: Detecting amount of memory? Hi, I trying to create a kickstart script on CentOS 5.5 that installs an RPM depending on the amount of memory in the box. For example if the box has >= 6GB memory then install RPM2 else install RPM1. With some Googling I now have the following snippet: %pre memKB=$(free|grep Mem|awk '{print$2}') memGB=$(echo "($memKB/1024)/1024"|bc) BIGMEM="" if [ "$memGB" -ge 6 ]; then BIGMEM="6GB" fi %post yum install foobar$BIGMEM-1.0-1.rpm The problem is that on a box with 6GB memory this snippet says it does not have 6GB mem: memKB = 6123716 KB memGB = 5 GB Anyone know what I am doing wrong? Thanks! Patrick _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From ohadlevy at gmail.com Tue Dec 21 13:00:11 2010 From: ohadlevy at gmail.com (Ohad Levy) Date: Tue, 21 Dec 2010 15:00:11 +0200 Subject: Detecting amount of memory? In-Reply-To: <4D109B1D.40500@puzzled.xs4all.nl> References: <4D109B1D.40500@puzzled.xs4all.nl> Message-ID: bc will round the value down, you need to change your calculation to round up or using a different scale. You would need to try it out, maybe something in the spirit of : echo "scale 1; $memKB/1024.... |bc if memGB -ge 5.5 ... Ohad On Tue, Dec 21, 2010 at 2:18 PM, Patrick Lists wrote: > Hi, > > I trying to create a kickstart script on CentOS 5.5 that installs an RPM > depending on the amount of memory in the box. For example if the box has > >= 6GB memory then install RPM2 else install RPM1. With some Googling I > now have the following snippet: > > %pre > memKB=$(free|grep Mem|awk '{print$2}') > memGB=$(echo "($memKB/1024)/1024"|bc) > BIGMEM="" > if [ "$memGB" -ge 6 ]; then > BIGMEM="6GB" > fi > > %post > yum install foobar$BIGMEM-1.0-1.rpm > > The problem is that on a box with 6GB memory this snippet says it does > not have 6GB mem: > > memKB = 6123716 KB > memGB = 5 GB > > Anyone know what I am doing wrong? > > Thanks! > Patrick > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From CallahanT at tessco.com Tue Dec 21 13:03:51 2010 From: CallahanT at tessco.com (Callahan, Tom) Date: Tue, 21 Dec 2010 08:03:51 -0500 Subject: Detecting amount of memory? In-Reply-To: <4D109B1D.40500@puzzled.xs4all.nl> Message-ID: You could do memGB=$(echo "($memKB/1024)/1000"|bc) That would seem to work based on your numbers On 12/21/10 7:18 AM, "Patrick Lists" wrote: > Hi, > > I trying to create a kickstart script on CentOS 5.5 that installs an RPM > depending on the amount of memory in the box. For example if the box has >> = 6GB memory then install RPM2 else install RPM1. With some Googling I > now have the following snippet: > > %pre > memKB=$(free|grep Mem|awk '{print$2}') > memGB=$(echo "($memKB/1024)/1024"|bc) > BIGMEM="" > if [ "$memGB" -ge 6 ]; then > BIGMEM="6GB" > fi > > %post > yum install foobar$BIGMEM-1.0-1.rpm > > The problem is that on a box with 6GB memory this snippet says it does > not have 6GB mem: > > memKB = 6123716 KB > memGB = 5 GB > > Anyone know what I am doing wrong? > > Thanks! > Patrick > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -- Tom Callahan Technology, Development and Services (TDS) 410-229-1361 Tel 410-229-1512 Fax CallahanT at TESSCO.com Visit TESSCO.com TESSCO Your Total Source? for Everything Wireless Network Infrastructure Equipment | Mobile Devices & Accessories | Installation, Test, Equipment & Supplies From CallahanT at tessco.com Tue Dec 21 13:05:44 2010 From: CallahanT at tessco.com (Callahan, Tom) Date: Tue, 21 Dec 2010 08:05:44 -0500 Subject: Detecting amount of memory? In-Reply-To: <4D109B1D.40500@puzzled.xs4all.nl> Message-ID: The trick is, 6123716 KB isn't actually 6GB, it's slightly less. That's why you are getting the results you are getting. http://www.google.com/search?q=6123716KB+in+GB On 12/21/10 7:18 AM, "Patrick Lists" wrote: > 6123716 -- Tom Callahan Technology, Development and Services (TDS) 410-229-1361 Tel 410-229-1512 Fax CallahanT at TESSCO.com Visit TESSCO.com TESSCO Your Total Source? for Everything Wireless Network Infrastructure Equipment | Mobile Devices & Accessories | Installation, Test, Equipment & Supplies From ks-list at puzzled.xs4all.nl Tue Dec 21 13:10:07 2010 From: ks-list at puzzled.xs4all.nl (Patrick Lists) Date: Tue, 21 Dec 2010 14:10:07 +0100 Subject: Detecting amount of memory? In-Reply-To: <591391867-1292935156-cardhu_decombobulator_blackberry.rim.net-1942212958-@bda2469.bisx.prod.on.blackberry> References: <4D109B1D.40500@puzzled.xs4all.nl> <591391867-1292935156-cardhu_decombobulator_blackberry.rim.net-1942212958-@bda2469.bisx.prod.on.blackberry> Message-ID: <4D10A72F.4020609@puzzled.xs4all.nl> On 12/21/2010 01:39 PM, jamesaharrisonuk at yahoo.co.uk wrote: > The kernel reports the amount of memory in /proc/meninfo . It shows the memory information in kilobytes. > James > Sent from my Verizon Wireless BlackBerry Thank you for your suggestion James. I tried it out and /proc/meminfo seems to give the same amount of memory as free: $ grep "MemTotal" /proc/meminfo | awk '{ print $2 }' 6123716 Fortunately I did find some awk voodoo that works. For the archives here it is. Mind the line breaks. Line 2,3,4 should be all one line. mem=$(free|grep Mem|awk '{print$2}') echo "$mem" | awk '{ n = $0; sub(/[0-9]+(.[0-9]+)?/, ""); split(n, a, $0); n=a[1]; if ($0 == "G" || $0 == "") { n *= 1024 }; if (/^kB?/) { n /= 1024 }; printf "%d\n", n }' # Intentionaly round to an integer This line gives 6144 on a box with 6GB mem. Regards, Patrick [snip] From ks-list at puzzled.xs4all.nl Tue Dec 21 13:29:41 2010 From: ks-list at puzzled.xs4all.nl (Patrick Lists) Date: Tue, 21 Dec 2010 14:29:41 +0100 Subject: Detecting amount of memory? In-Reply-To: References: Message-ID: <4D10ABC5.20309@puzzled.xs4all.nl> On 12/21/2010 02:05 PM, Callahan, Tom wrote: > The trick is, 6123716 KB isn't actually 6GB, it's slightly less. That's why > you are getting the results you are getting. > > http://www.google.com/search?q=6123716KB+in+GB > > > On 12/21/10 7:18 AM, "Patrick Lists" wrote: > >> 6123716 Yup see that now. Still I wonder why free says MemTotal is less than 6GB when the box has 6GB. That's the interesting part. I seem to be missing something. Where did those last few bytes go? Or could this be a case of GB versus GiB. Strangely this awk line does work but I have no awk foo so this is like dark arts to me (line 2,3,4 should all be on one line): mem=$(free|grep Mem|awk '{print$2}') echo "$mem" | awk '{ n = $0; sub(/[0-9]+(.[0-9]+)?/, ""); split(n, a, $0); n=a[1]; if ($0 == "G" || $0 == "") { n *= 1024 }; if (/^kB?/) { n /= 1024 }; printf "%d\n", n }' # Intentionally rounded to an integer Regards, Patrick From cjk at techma.com Tue Dec 21 13:41:06 2010 From: cjk at techma.com (Corey Kovacs) Date: Tue, 21 Dec 2010 13:41:06 +0000 Subject: Detecting amount of memory? In-Reply-To: References: <4D109B1D.40500@puzzled.xs4all.nl> Message-ID: <0ADCB4FF-6500-454B-ABD9-2F6E1F4BD596@techma.com> Another way to do this is grep for the zone info from dmesg and do the calcs on that info. You need to account for low mem as well which is usually about 800+kb. Also dma etc. I can't remember if dmidecode is available during a kickstart but you could also run that and grep for the dimms (Size: ) should let you see what I mean assuming your not building VMs. Sent from my iPod On Dec 21, 2010, at 1:00 PM, Ohad Levy wrote: > bc will round the value down, you need to change your calculation to round up or using a different scale. > > You would need to try it out, maybe something in the spirit of : > > echo "scale 1; $memKB/1024.... |bc > if memGB -ge 5.5 ... > > Ohad > > > On Tue, Dec 21, 2010 at 2:18 PM, Patrick Lists wrote: > Hi, > > I trying to create a kickstart script on CentOS 5.5 that installs an RPM > depending on the amount of memory in the box. For example if the box has > >= 6GB memory then install RPM2 else install RPM1. With some Googling I > now have the following snippet: > > %pre > memKB=$(free|grep Mem|awk '{print$2}') > memGB=$(echo "($memKB/1024)/1024"|bc) > BIGMEM="" > if [ "$memGB" -ge 6 ]; then > BIGMEM="6GB" > fi > > %post > yum install foobar$BIGMEM-1.0-1.rpm > > The problem is that on a box with 6GB memory this snippet says it does > not have 6GB mem: > > memKB = 6123716 KB > memGB = 5 GB > > Anyone know what I am doing wrong? > > Thanks! > Patrick > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list -------------- next part -------------- An HTML attachment was scrubbed... URL: From clumens at redhat.com Tue Dec 21 14:00:05 2010 From: clumens at redhat.com (Chris Lumens) Date: Tue, 21 Dec 2010 09:00:05 -0500 Subject: Anaconda stage 1 and ssl In-Reply-To: <4325_1292887138_oBKNIqQP008913_4D0FE4BD.1040604@rackspace.com> References: <4325_1292887138_oBKNIqQP008913_4D0FE4BD.1040604@rackspace.com> Message-ID: <20101221140005.GK30483@exeter.usersys.redhat.com> > I am attempting to host my ks.cfg file on an SSL enabled web server. > Unfortunately, the installer does not seem to have http-ssl support. > Am I missing an option? Or is this the sad truth? > > The user agent is reported as: "anaconda/11.1.2.113" HTTPS was not supported in anaconda until anaconda-13.0. - Chris From hhoffman at ip-solutions.net Tue Dec 21 14:05:29 2010 From: hhoffman at ip-solutions.net (=?utf-8?B?SGFycnkgSG9mZm1hbg==?=) Date: Tue, 21 Dec 2010 09:05:29 -0500 Subject: =?utf-8?B?UmU6IEFuYWNvbmRhIHN0YWdlIDEgYW5kIHNzbA==?= Message-ID: <20101221140514.444418A29@b-sasl-quonix.pobox.com> This is the sad truth -- sent from my phone ----- Reply message ----- From: "Jared Rodriguez" Date: Mon, Dec 20, 2010 6:20 pm Subject: Anaconda stage 1 and ssl To: "Discussion list about Kickstart" I am attempting to host my ks.cfg file on an SSL enabled web server. Unfortunately, the installer does not seem to have http-ssl support. Am I missing an option? Or is this the sad truth? The user agent is reported as: "anaconda/11.1.2.113" Jared Confidentiality Notice: This e-mail message (including any attached or embedded documents) is intended for the exclusive and confidential use of the individual or entity to which this message is addressed, and unless otherwise expressly indicated, is confidential and privileged information of Rackspace. Any dissemination, distribution or copying of the enclosed material is prohibited. If you receive this transmission in error, please notify us immediately by e-mail at abuse at rackspace.com, and delete the original message. Your cooperation is appreciated. _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list -------------- next part -------------- An HTML attachment was scrubbed... URL: From jamesaharrisonuk at yahoo.co.uk Tue Dec 21 15:14:23 2010 From: jamesaharrisonuk at yahoo.co.uk (James Harrison) Date: Tue, 21 Dec 2010 15:14:23 +0000 (GMT) Subject: Detecting amount of memory? In-Reply-To: <4D10A72F.4020609@puzzled.xs4all.nl> References: <4D109B1D.40500@puzzled.xs4all.nl> <591391867-1292935156-cardhu_decombobulator_blackberry.rim.net-1942212958-@bda2469.bisx.prod.on.blackberry> <4D10A72F.4020609@puzzled.xs4all.nl> Message-ID: <225152.46500.qm@web29505.mail.ird.yahoo.com> cat /proc/meminfo | awk '/MemTotal:/ { m = $2 f = m /= 1024 # Mb f /= 1024 # Gb } END {printf "%0.f\n", f }' on my machine I get 16 = 16 Gb ________________________________ From: Patrick Lists To: jamesaharrisonuk at yahoo.co.uk; Discussion list about Kickstart Sent: Tue, December 21, 2010 8:10:07 AM Subject: Re: Detecting amount of memory? On 12/21/2010 01:39 PM, jamesaharrisonuk at yahoo.co.uk wrote: > The kernel reports the amount of memory in /proc/meninfo . It shows the memory >information in kilobytes. > James > Sent from my Verizon Wireless BlackBerry Thank you for your suggestion James. I tried it out and /proc/meminfo seems to give the same amount of memory as free: $ grep "MemTotal" /proc/meminfo | awk '{ print $2 }' 6123716 Fortunately I did find some awk voodoo that works. For the archives here it is. Mind the line breaks. Line 2,3,4 should be all one line. mem=$(free|grep Mem|awk '{print$2}') echo "$mem" | awk '{ n = $0; sub(/[0-9]+(.[0-9]+)?/, ""); split(n, a, $0); n=a[1]; if ($0 == "G" || $0 == "") { n *= 1024 }; if (/^kB?/) { n /= 1024 }; printf "%d\n", n }' # Intentionaly round to an integer This line gives 6144 on a box with 6GB mem. Regards, Patrick [snip] -------------- next part -------------- An HTML attachment was scrubbed... URL: From clumens at redhat.com Tue Dec 21 17:15:19 2010 From: clumens at redhat.com (Chris Lumens) Date: Tue, 21 Dec 2010 12:15:19 -0500 Subject: RHEL 6 Anaconda Error - AttributeError In-Reply-To: References: <20101213155222.GL30483@exeter.usersys.redhat.com> Message-ID: <20101221171518.GM30483@exeter.usersys.redhat.com> > Here's a link to the anaconda-tb-* ... > > http://www.aqr.aero/files/anaconda-tb-tTx5Tj Hm, that wasn't very enlightening. This is one of those bugs where a lot of people should be hitting it, but this is the first I have heard of it. Can you think of anything unusual you did during install? Did you run media check? Is it at all reproducible? - Chris From jnaikwadi at asterdata.com Tue Dec 21 19:47:37 2010 From: jnaikwadi at asterdata.com (Jay Naikwadi) Date: Tue, 21 Dec 2010 11:47:37 -0800 Subject: RAID chunk size configuration from kickstart In-Reply-To: References: Message-ID: <4D110459.5050609@asterdata.com> Well in my case the nodes did have exact same config so that is not issue. But if kickstart command itself supported additional parameters it would have been lot simpler to specify chunk size. As in other cases we need to worry about different devices names(sd*/cciss*) etc and do all of this stuff in %post section to set the chunk size. But just to confirm , so there is no way this can be done using kickstart 'raid' command and has to be done using tools like mdadm in %pre / %post sections ? Thanks, Jay kickstart-list-request at redhat.com wrote: > Send Kickstart-list mailing list submissions to > kickstart-list at redhat.com > > To subscribe or unsubscribe via the World Wide Web, visit > https://www.redhat.com/mailman/listinfo/kickstart-list > or, via email, send a message with subject or body 'help' to > kickstart-list-request at redhat.com > > You can reach the person managing the list at > kickstart-list-owner at redhat.com > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Kickstart-list digest..." > > ------------------------------------------------------------------------ > > Subject: > RE: RAID chunk size configuration from kickstart > From: > "Kovacs, Corey J." > Date: > Fri, 17 Dec 2010 23:04:28 -0800 > To: > Discussion list about Kickstart > > To: > Discussion list about Kickstart > > > > > > > > > As far as I know, not for the root fs which "normally" shouldn't be a > problem since setting the chunk size is a performance tuning task and > you typically don't do high load tasks against your root fs (Well I > don't anyway). So, I'd suggest simply building a small root fs using > the normal kick start methods, then elsewhere in your post section, > manually create the raid device and file-system where you need to set > the chunk size. Also, this is sort of predicated by all of your nodes > are exactly the same and have exactly the same (or very similar) > workload. -C > ------------------------------------------------------------------------ > *From:* kickstart-list-bounces at redhat.com > [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi > [jnaikwadi at asterdata.com] *Sent:* Wednesday, December 15, 2010 2:10 PM > *To:* kickstart-list at redhat.com *Subject:* Re: RAID chunk size > configuration from kickstart > Jay Naikwadi wrote: >> Hi , I am trying to deploy software raid setup using kickstart and >> was wondering if I can specify chunk-size as parameter in kickstart >> file.I am using RHEL 6. I didn't see any documentation for this >> parameter in kickstart. The default installation seems to be using >> 512K as chunk size whereas I would like to set it to 8MB . This is >> new device configuration and there is nothing configured on system >> before this. If kickstart does not allow setting this parameter , Is >> there another way in %pre , %post sections where we can change it ? >> How else this parameter could be set for '/' , '/boot/' >> partitions on clean install ? Thanks and regards, Jay > Any suggestions about how this could be done ? -Jay > > From cjk at techma.com Tue Dec 21 20:16:02 2010 From: cjk at techma.com (Kovacs, Corey J.) Date: Tue, 21 Dec 2010 20:16:02 +0000 Subject: RAID chunk size configuration from kickstart In-Reply-To: <4D110459.5050609@asterdata.com> References: , <4D110459.5050609@asterdata.com> Message-ID: Just curious, why do you need an 8M chunk size? What sort of devices are you writing to? -C ________________________________________ From: kickstart-list-bounces at redhat.com [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi [jnaikwadi at asterdata.com] Sent: Tuesday, December 21, 2010 2:47 PM To: kickstart-list at redhat.com Subject: RE: RAID chunk size configuration from kickstart Well in my case the nodes did have exact same config so that is not issue. But if kickstart command itself supported additional parameters it would have been lot simpler to specify chunk size. As in other cases we need to worry about different devices names(sd*/cciss*) etc and do all of this stuff in %post section to set the chunk size. But just to confirm , so there is no way this can be done using kickstart 'raid' command and has to be done using tools like mdadm in %pre / %post sections ? Thanks, Jay kickstart-list-request at redhat.com wrote: > Send Kickstart-list mailing list submissions to > kickstart-list at redhat.com > > To subscribe or unsubscribe via the World Wide Web, visit > https://www.redhat.com/mailman/listinfo/kickstart-list > or, via email, send a message with subject or body 'help' to > kickstart-list-request at redhat.com > > You can reach the person managing the list at > kickstart-list-owner at redhat.com > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Kickstart-list digest..." > > ------------------------------------------------------------------------ > > Subject: > RE: RAID chunk size configuration from kickstart > From: > "Kovacs, Corey J." > Date: > Fri, 17 Dec 2010 23:04:28 -0800 > To: > Discussion list about Kickstart > > To: > Discussion list about Kickstart > > > > > > > > > As far as I know, not for the root fs which "normally" shouldn't be a > problem since setting the chunk size is a performance tuning task and > you typically don't do high load tasks against your root fs (Well I > don't anyway). So, I'd suggest simply building a small root fs using > the normal kick start methods, then elsewhere in your post section, > manually create the raid device and file-system where you need to set > the chunk size. Also, this is sort of predicated by all of your nodes > are exactly the same and have exactly the same (or very similar) > workload. -C > ------------------------------------------------------------------------ > *From:* kickstart-list-bounces at redhat.com > [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi > [jnaikwadi at asterdata.com] *Sent:* Wednesday, December 15, 2010 2:10 PM > *To:* kickstart-list at redhat.com *Subject:* Re: RAID chunk size > configuration from kickstart > Jay Naikwadi wrote: >> Hi , I am trying to deploy software raid setup using kickstart and >> was wondering if I can specify chunk-size as parameter in kickstart >> file.I am using RHEL 6. I didn't see any documentation for this >> parameter in kickstart. The default installation seems to be using >> 512K as chunk size whereas I would like to set it to 8MB . This is >> new device configuration and there is nothing configured on system >> before this. If kickstart does not allow setting this parameter , Is >> there another way in %pre , %post sections where we can change it ? >> How else this parameter could be set for '/' , '/boot/' >> partitions on clean install ? Thanks and regards, Jay > Any suggestions about how this could be done ? -Jay > > _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From andy.speagle at wichita.edu Wed Dec 22 16:09:57 2010 From: andy.speagle at wichita.edu (Speagle, Andy) Date: Wed, 22 Dec 2010 10:09:57 -0600 Subject: RHEL 6 Anaconda Error - AttributeError In-Reply-To: <20101221171518.GM30483@exeter.usersys.redhat.com> References: <20101213155222.GL30483@exeter.usersys.redhat.com> <20101221171518.GM30483@exeter.usersys.redhat.com> Message-ID: > > Here's a link to the anaconda-tb-* ... > > > > http://www.aqr.aero/files/anaconda-tb-tTx5Tj > > Hm, that wasn't very enlightening. This is one of those bugs where a lot > of people should be hitting it, but this is the first I have heard of it. > Can you think of anything unusual you did during install? Did you run > media check? Is it at all reproducible? Well, keep in mind this is a kickstart generated via Spacewalk v1.2 ... I cloned the majority of it from a RHEL 5 kickstart. I'm not certain if there are new directives I need to be considering. Here's the relevant generated kickstart... Perhaps there's something you can readily identify. # Kickstart config file generated by Spacewalk Config Management # Profile Label : rhel-x86_64-server-6-0 # Date Created : 2010-11-29 21:27:30.0 install text network --device=eth0 --bootproto=dhcp --noipv6 url --url http://spacewalk.wichita.edu/ks/dist/rhel-x86_64-server-6-0 lang en_US keyboard us zerombr yes clearpart --all --drives=sda --initlabel bootloader --location mbr --driveorder=sda --append="rhgb quiet" timezone --utc America/Chicago auth --enablemd5 --enableshadow rootpw --iscrypted selinux --disabled reboot firewall --enabled --port=22:tcp skipx firstboot --disabled part /boot --fstype ext4 --size=200 --ondisk=sda part pv.6 --size=100 --grow --ondisk=sda volgroup rootvg --pesize=32768 pv.6 logvol swap --fstype=swap --name=swap --vgname=rootvg --size=2048 logvol /var --fstype=ext4 --name=var --vgname=rootvg --size=2048 logvol /var/tmp --fstype=ext4 --name=vartmp --vgname=rootvg --size=1024 logvol /home --fstype=ext4 --name=home --vgname=rootvg --size=2048 logvol /tmp --fstype=ext4 --name=tmp --vgname=rootvg --size=1024 logvol /opt --fstype=ext4 --name=opt --vgname=rootvg --size=2048 logvol / --fstype=ext4 --name=root --vgname=rootvg --size=4096 %packages @core @base @java lvm2 gcc gcc-c++ net-snmp sysstat strace ntp screen compat-libstdc++-33 -NetworkManager* -bluez* -yum-updatesd %end Thanks, Andy From jnaikwadi at asterdata.com Thu Dec 23 00:23:12 2010 From: jnaikwadi at asterdata.com (Jay Naikwadi) Date: Wed, 22 Dec 2010 16:23:12 -0800 Subject: RAID chunk size configuration from kickstart In-Reply-To: References: Message-ID: <4D129670.3090207@asterdata.com> >>Just curious, why do you need an 8M chunk size? What sort of devices are you writing to? >>-C Well the box is normal dell C2100 box with (LSI MegaRAID SAS 9260-8i w/ 512 MB cache 12x 450GB, 15K SAS). I am playing with different RAID configuration to see what numbers I get for pure block I/O performance. But looks like I will have to get this done in %post section. Just wondering though why does not kickstart allow setting chunk size directly if underlying utilities support this parameter ? -Jay kickstart-list-request at redhat.com wrote: > Send Kickstart-list mailing list submissions to > kickstart-list at redhat.com > > To subscribe or unsubscribe via the World Wide Web, visit > https://www.redhat.com/mailman/listinfo/kickstart-list > or, via email, send a message with subject or body 'help' to > kickstart-list-request at redhat.com > > You can reach the person managing the list at > kickstart-list-owner at redhat.com > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Kickstart-list digest..." > > ------------------------------------------------------------------------ > > Subject: > Re: RHEL 6 Anaconda Error - AttributeError > From: > Chris Lumens > Date: > Tue, 21 Dec 2010 09:15:19 -0800 > To: > "kickstart-list at redhat.com" > > To: > "kickstart-list at redhat.com" > > > >> Here's a link to the anaconda-tb-* ... >> >> http://www.aqr.aero/files/anaconda-tb-tTx5Tj > > Hm, that wasn't very enlightening. This is one of those bugs where a > lot of people should be hitting it, but this is the first I have heard > of it. Can you think of anything unusual you did during install? Did > you run media check? Is it at all reproducible? > > - Chris > > > > ------------------------------------------------------------------------ > > Subject: > RE: RAID chunk size configuration from kickstart > From: > Jay Naikwadi > Date: > Tue, 21 Dec 2010 11:47:37 -0800 > To: > "kickstart-list at redhat.com" > > To: > "kickstart-list at redhat.com" > > > > Well in my case the nodes did have exact same config so that is not issue. > But if kickstart command itself supported additional parameters it would > have been lot simpler to specify chunk size. As in other cases we need > to worry about different devices names(sd*/cciss*) etc and do all of > this stuff in %post section to set the chunk size. > > But just to confirm , so there is no way this can be done using > kickstart 'raid' command and has to be done using tools like mdadm in > %pre / %post sections ? > > Thanks, > Jay > > kickstart-list-request at redhat.com wrote: >> Send Kickstart-list mailing list submissions to >> kickstart-list at redhat.com >> >> To subscribe or unsubscribe via the World Wide Web, visit >> https://www.redhat.com/mailman/listinfo/kickstart-list >> or, via email, send a message with subject or body 'help' to >> kickstart-list-request at redhat.com >> >> You can reach the person managing the list at >> kickstart-list-owner at redhat.com >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of Kickstart-list digest..." >> >> ------------------------------------------------------------------------ >> >> Subject: >> RE: RAID chunk size configuration from kickstart >> From: >> "Kovacs, Corey J." >> Date: >> Fri, 17 Dec 2010 23:04:28 -0800 >> To: >> Discussion list about Kickstart >> >> To: >> Discussion list about Kickstart >> >> >> >> >> >> >> >> >> As far as I know, not for the root fs which "normally" shouldn't be a >> problem since setting the chunk size is a performance tuning task and >> you typically don't do high load tasks against your root fs (Well I >> don't anyway). So, I'd suggest simply building a small root fs using >> the normal kick start methods, then elsewhere in your post section, >> manually create the raid device and file-system where you need to set >> the chunk size. Also, this is sort of predicated by all of your nodes >> are exactly the same and have exactly the same (or very similar) >> workload. -C >> ------------------------------------------------------------------------ >> *From:* kickstart-list-bounces at redhat.com >> [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi >> [jnaikwadi at asterdata.com] *Sent:* Wednesday, December 15, 2010 2:10 PM >> *To:* kickstart-list at redhat.com *Subject:* Re: RAID chunk size >> configuration from kickstart >> Jay Naikwadi wrote: >>> Hi , I am trying to deploy software raid setup using kickstart and >>> was wondering if I can specify chunk-size as parameter in kickstart >>> file.I am using RHEL 6. I didn't see any documentation for this >>> parameter in kickstart. The default installation seems to be using >>> 512K as chunk size whereas I would like to set it to 8MB . This is >>> new device configuration and there is nothing configured on system >>> before this. If kickstart does not allow setting this parameter , Is >>> there another way in %pre , %post sections where we can change it ? >>> How else this parameter could be set for '/' , '/boot/' >>> partitions on clean install ? Thanks and regards, Jay >> Any suggestions about how this could be done ? -Jay >> >> > > > > ------------------------------------------------------------------------ > > Subject: > RE: RAID chunk size configuration from kickstart > From: > "Kovacs, Corey J." > Date: > Tue, 21 Dec 2010 12:16:02 -0800 > To: > Discussion list about Kickstart > > To: > Discussion list about Kickstart > > > > Just curious, why do you need an 8M chunk size? What sort of devices are you writing to? > > -C > ________________________________________ > From: kickstart-list-bounces at redhat.com [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi [jnaikwadi at asterdata.com] > Sent: Tuesday, December 21, 2010 2:47 PM > To: kickstart-list at redhat.com > Subject: RE: RAID chunk size configuration from kickstart > > Well in my case the nodes did have exact same config so that is not issue. > But if kickstart command itself supported additional parameters it would > have been lot simpler to specify chunk size. As in other cases we need > to worry about different devices names(sd*/cciss*) etc and do all of > this stuff in %post section to set the chunk size. > > But just to confirm , so there is no way this can be done using > kickstart 'raid' command and has to be done using tools like mdadm in > %pre / %post sections ? > > Thanks, > Jay > > kickstart-list-request at redhat.com wrote: >> Send Kickstart-list mailing list submissions to >> kickstart-list at redhat.com >> >> To subscribe or unsubscribe via the World Wide Web, visit >> https://www.redhat.com/mailman/listinfo/kickstart-list >> or, via email, send a message with subject or body 'help' to >> kickstart-list-request at redhat.com >> >> You can reach the person managing the list at >> kickstart-list-owner at redhat.com >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of Kickstart-list digest..." >> >> ------------------------------------------------------------------------ >> >> Subject: >> RE: RAID chunk size configuration from kickstart >> From: >> "Kovacs, Corey J." >> Date: >> Fri, 17 Dec 2010 23:04:28 -0800 >> To: >> Discussion list about Kickstart >> >> To: >> Discussion list about Kickstart >> >> >> >> >> >> >> >> >> As far as I know, not for the root fs which "normally" shouldn't be a >> problem since setting the chunk size is a performance tuning task and >> you typically don't do high load tasks against your root fs (Well I >> don't anyway). So, I'd suggest simply building a small root fs using >> the normal kick start methods, then elsewhere in your post section, >> manually create the raid device and file-system where you need to set >> the chunk size. Also, this is sort of predicated by all of your nodes >> are exactly the same and have exactly the same (or very similar) >> workload. -C >> ------------------------------------------------------------------------ >> *From:* kickstart-list-bounces at redhat.com >> [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi >> [jnaikwadi at asterdata.com] *Sent:* Wednesday, December 15, 2010 2:10 PM >> *To:* kickstart-list at redhat.com *Subject:* Re: RAID chunk size >> configuration from kickstart >> Jay Naikwadi wrote: >>> Hi , I am trying to deploy software raid setup using kickstart and >>> was wondering if I can specify chunk-size as parameter in kickstart >>> file.I am using RHEL 6. I didn't see any documentation for this >>> parameter in kickstart. The default installation seems to be using >>> 512K as chunk size whereas I would like to set it to 8MB . This is >>> new device configuration and there is nothing configured on system >>> before this. If kickstart does not allow setting this parameter , Is >>> there another way in %pre , %post sections where we can change it ? >>> How else this parameter could be set for '/' , '/boot/' >>> partitions on clean install ? Thanks and regards, Jay >> Any suggestions about how this could be done ? -Jay >> >> > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > > > ------------------------------------------------------------------------ > > Subject: > RE: RHEL 6 Anaconda Error - AttributeError > From: > "Speagle, Andy" > Date: > Wed, 22 Dec 2010 08:09:57 -0800 > To: > Discussion list about Kickstart > > To: > Discussion list about Kickstart > > > >>> Here's a link to the anaconda-tb-* ... >>> >>> http://www.aqr.aero/files/anaconda-tb-tTx5Tj >> Hm, that wasn't very enlightening. This is one of those bugs where a lot >> of people should be hitting it, but this is the first I have heard of it. >> Can you think of anything unusual you did during install? Did you run >> media check? Is it at all reproducible? > > Well, keep in mind this is a kickstart generated via Spacewalk v1.2 ... I cloned the majority of it from a RHEL 5 kickstart. I'm not certain if there are new directives I need to be considering. Here's the relevant generated kickstart... Perhaps there's something you can readily identify. > > # Kickstart config file generated by Spacewalk Config Management > # Profile Label : rhel-x86_64-server-6-0 > # Date Created : 2010-11-29 21:27:30.0 > > install > text > network --device=eth0 --bootproto=dhcp --noipv6 > url --url http://spacewalk.wichita.edu/ks/dist/rhel-x86_64-server-6-0 > lang en_US > keyboard us > zerombr yes > clearpart --all --drives=sda --initlabel > bootloader --location mbr --driveorder=sda --append="rhgb quiet" > timezone --utc America/Chicago > auth --enablemd5 --enableshadow > rootpw --iscrypted > selinux --disabled > reboot > firewall --enabled --port=22:tcp > skipx > firstboot --disabled > part /boot --fstype ext4 --size=200 --ondisk=sda > part pv.6 --size=100 --grow --ondisk=sda > volgroup rootvg --pesize=32768 pv.6 > logvol swap --fstype=swap --name=swap --vgname=rootvg --size=2048 > logvol /var --fstype=ext4 --name=var --vgname=rootvg --size=2048 > logvol /var/tmp --fstype=ext4 --name=vartmp --vgname=rootvg --size=1024 > logvol /home --fstype=ext4 --name=home --vgname=rootvg --size=2048 > logvol /tmp --fstype=ext4 --name=tmp --vgname=rootvg --size=1024 > logvol /opt --fstype=ext4 --name=opt --vgname=rootvg --size=2048 > logvol / --fstype=ext4 --name=root --vgname=rootvg --size=4096 > > %packages > @core > @base > @java > lvm2 > gcc > gcc-c++ > net-snmp > sysstat > strace > ntp > screen > compat-libstdc++-33 > -NetworkManager* > -bluez* > -yum-updatesd > %end > > > Thanks, > > Andy > > From cjk at techma.com Thu Dec 23 07:00:49 2010 From: cjk at techma.com (Kovacs, Corey J.) Date: Thu, 23 Dec 2010 07:00:49 +0000 Subject: RAID chunk size configuration from kickstart In-Reply-To: <4D129670.3090207@asterdata.com> References: , <4D129670.3090207@asterdata.com> Message-ID: My guess is that it's usually a setting used based on a known metric of I/O load. Only when you know exactly what the average load for the particular set of disks is for a particular usage pattern, can you begin define what the chunk size is. Experimentation is of course the only real way to find the right value but once you do, and you combine that with the stride settings for the file system you lay on top of the raid it will help. That all said, I guess it'd be nice to have as an option since one might have a known set of systems already doing the same thing. Anyway, good luck. -C ________________________________________ From: kickstart-list-bounces at redhat.com [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi [jnaikwadi at asterdata.com] Sent: Wednesday, December 22, 2010 7:23 PM To: kickstart-list at redhat.com Subject: RE: RAID chunk size configuration from kickstart >>Just curious, why do you need an 8M chunk size? What sort of devices are you writing to? >>-C Well the box is normal dell C2100 box with (LSI MegaRAID SAS 9260-8i w/ 512 MB cache 12x 450GB, 15K SAS). I am playing with different RAID configuration to see what numbers I get for pure block I/O performance. But looks like I will have to get this done in %post section. Just wondering though why does not kickstart allow setting chunk size directly if underlying utilities support this parameter ? -Jay kickstart-list-request at redhat.com wrote: > Send Kickstart-list mailing list submissions to > kickstart-list at redhat.com > > To subscribe or unsubscribe via the World Wide Web, visit > https://www.redhat.com/mailman/listinfo/kickstart-list > or, via email, send a message with subject or body 'help' to > kickstart-list-request at redhat.com > > You can reach the person managing the list at > kickstart-list-owner at redhat.com > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Kickstart-list digest..." > > ------------------------------------------------------------------------ > > Subject: > Re: RHEL 6 Anaconda Error - AttributeError > From: > Chris Lumens > Date: > Tue, 21 Dec 2010 09:15:19 -0800 > To: > "kickstart-list at redhat.com" > > To: > "kickstart-list at redhat.com" > > > >> Here's a link to the anaconda-tb-* ... >> >> http://www.aqr.aero/files/anaconda-tb-tTx5Tj > > Hm, that wasn't very enlightening. This is one of those bugs where a > lot of people should be hitting it, but this is the first I have heard > of it. Can you think of anything unusual you did during install? Did > you run media check? Is it at all reproducible? > > - Chris > > > > ------------------------------------------------------------------------ > > Subject: > RE: RAID chunk size configuration from kickstart > From: > Jay Naikwadi > Date: > Tue, 21 Dec 2010 11:47:37 -0800 > To: > "kickstart-list at redhat.com" > > To: > "kickstart-list at redhat.com" > > > > Well in my case the nodes did have exact same config so that is not issue. > But if kickstart command itself supported additional parameters it would > have been lot simpler to specify chunk size. As in other cases we need > to worry about different devices names(sd*/cciss*) etc and do all of > this stuff in %post section to set the chunk size. > > But just to confirm , so there is no way this can be done using > kickstart 'raid' command and has to be done using tools like mdadm in > %pre / %post sections ? > > Thanks, > Jay > > kickstart-list-request at redhat.com wrote: >> Send Kickstart-list mailing list submissions to >> kickstart-list at redhat.com >> >> To subscribe or unsubscribe via the World Wide Web, visit >> https://www.redhat.com/mailman/listinfo/kickstart-list >> or, via email, send a message with subject or body 'help' to >> kickstart-list-request at redhat.com >> >> You can reach the person managing the list at >> kickstart-list-owner at redhat.com >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of Kickstart-list digest..." >> >> ------------------------------------------------------------------------ >> >> Subject: >> RE: RAID chunk size configuration from kickstart >> From: >> "Kovacs, Corey J." >> Date: >> Fri, 17 Dec 2010 23:04:28 -0800 >> To: >> Discussion list about Kickstart >> >> To: >> Discussion list about Kickstart >> >> >> >> >> >> >> >> >> As far as I know, not for the root fs which "normally" shouldn't be a >> problem since setting the chunk size is a performance tuning task and >> you typically don't do high load tasks against your root fs (Well I >> don't anyway). So, I'd suggest simply building a small root fs using >> the normal kick start methods, then elsewhere in your post section, >> manually create the raid device and file-system where you need to set >> the chunk size. Also, this is sort of predicated by all of your nodes >> are exactly the same and have exactly the same (or very similar) >> workload. -C >> ------------------------------------------------------------------------ >> *From:* kickstart-list-bounces at redhat.com >> [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi >> [jnaikwadi at asterdata.com] *Sent:* Wednesday, December 15, 2010 2:10 PM >> *To:* kickstart-list at redhat.com *Subject:* Re: RAID chunk size >> configuration from kickstart >> Jay Naikwadi wrote: >>> Hi , I am trying to deploy software raid setup using kickstart and >>> was wondering if I can specify chunk-size as parameter in kickstart >>> file.I am using RHEL 6. I didn't see any documentation for this >>> parameter in kickstart. The default installation seems to be using >>> 512K as chunk size whereas I would like to set it to 8MB . This is >>> new device configuration and there is nothing configured on system >>> before this. If kickstart does not allow setting this parameter , Is >>> there another way in %pre , %post sections where we can change it ? >>> How else this parameter could be set for '/' , '/boot/' >>> partitions on clean install ? Thanks and regards, Jay >> Any suggestions about how this could be done ? -Jay >> >> > > > > ------------------------------------------------------------------------ > > Subject: > RE: RAID chunk size configuration from kickstart > From: > "Kovacs, Corey J." > Date: > Tue, 21 Dec 2010 12:16:02 -0800 > To: > Discussion list about Kickstart > > To: > Discussion list about Kickstart > > > > Just curious, why do you need an 8M chunk size? What sort of devices are you writing to? > > -C > ________________________________________ > From: kickstart-list-bounces at redhat.com [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi [jnaikwadi at asterdata.com] > Sent: Tuesday, December 21, 2010 2:47 PM > To: kickstart-list at redhat.com > Subject: RE: RAID chunk size configuration from kickstart > > Well in my case the nodes did have exact same config so that is not issue. > But if kickstart command itself supported additional parameters it would > have been lot simpler to specify chunk size. As in other cases we need > to worry about different devices names(sd*/cciss*) etc and do all of > this stuff in %post section to set the chunk size. > > But just to confirm , so there is no way this can be done using > kickstart 'raid' command and has to be done using tools like mdadm in > %pre / %post sections ? > > Thanks, > Jay > > kickstart-list-request at redhat.com wrote: >> Send Kickstart-list mailing list submissions to >> kickstart-list at redhat.com >> >> To subscribe or unsubscribe via the World Wide Web, visit >> https://www.redhat.com/mailman/listinfo/kickstart-list >> or, via email, send a message with subject or body 'help' to >> kickstart-list-request at redhat.com >> >> You can reach the person managing the list at >> kickstart-list-owner at redhat.com >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of Kickstart-list digest..." >> >> ------------------------------------------------------------------------ >> >> Subject: >> RE: RAID chunk size configuration from kickstart >> From: >> "Kovacs, Corey J." >> Date: >> Fri, 17 Dec 2010 23:04:28 -0800 >> To: >> Discussion list about Kickstart >> >> To: >> Discussion list about Kickstart >> >> >> >> >> >> >> >> >> As far as I know, not for the root fs which "normally" shouldn't be a >> problem since setting the chunk size is a performance tuning task and >> you typically don't do high load tasks against your root fs (Well I >> don't anyway). So, I'd suggest simply building a small root fs using >> the normal kick start methods, then elsewhere in your post section, >> manually create the raid device and file-system where you need to set >> the chunk size. Also, this is sort of predicated by all of your nodes >> are exactly the same and have exactly the same (or very similar) >> workload. -C >> ------------------------------------------------------------------------ >> *From:* kickstart-list-bounces at redhat.com >> [kickstart-list-bounces at redhat.com] on behalf of Jay Naikwadi >> [jnaikwadi at asterdata.com] *Sent:* Wednesday, December 15, 2010 2:10 PM >> *To:* kickstart-list at redhat.com *Subject:* Re: RAID chunk size >> configuration from kickstart >> Jay Naikwadi wrote: >>> Hi , I am trying to deploy software raid setup using kickstart and >>> was wondering if I can specify chunk-size as parameter in kickstart >>> file.I am using RHEL 6. I didn't see any documentation for this >>> parameter in kickstart. The default installation seems to be using >>> 512K as chunk size whereas I would like to set it to 8MB . This is >>> new device configuration and there is nothing configured on system >>> before this. If kickstart does not allow setting this parameter , Is >>> there another way in %pre , %post sections where we can change it ? >>> How else this parameter could be set for '/' , '/boot/' >>> partitions on clean install ? Thanks and regards, Jay >> Any suggestions about how this could be done ? -Jay >> >> > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > > > ------------------------------------------------------------------------ > > Subject: > RE: RHEL 6 Anaconda Error - AttributeError > From: > "Speagle, Andy" > Date: > Wed, 22 Dec 2010 08:09:57 -0800 > To: > Discussion list about Kickstart > > To: > Discussion list about Kickstart > > > >>> Here's a link to the anaconda-tb-* ... >>> >>> http://www.aqr.aero/files/anaconda-tb-tTx5Tj >> Hm, that wasn't very enlightening. This is one of those bugs where a lot >> of people should be hitting it, but this is the first I have heard of it. >> Can you think of anything unusual you did during install? Did you run >> media check? Is it at all reproducible? > > Well, keep in mind this is a kickstart generated via Spacewalk v1.2 ... I cloned the majority of it from a RHEL 5 kickstart. I'm not certain if there are new directives I need to be considering. Here's the relevant generated kickstart... Perhaps there's something you can readily identify. > > # Kickstart config file generated by Spacewalk Config Management > # Profile Label : rhel-x86_64-server-6-0 > # Date Created : 2010-11-29 21:27:30.0 > > install > text > network --device=eth0 --bootproto=dhcp --noipv6 > url --url http://spacewalk.wichita.edu/ks/dist/rhel-x86_64-server-6-0 > lang en_US > keyboard us > zerombr yes > clearpart --all --drives=sda --initlabel > bootloader --location mbr --driveorder=sda --append="rhgb quiet" > timezone --utc America/Chicago > auth --enablemd5 --enableshadow > rootpw --iscrypted > selinux --disabled > reboot > firewall --enabled --port=22:tcp > skipx > firstboot --disabled > part /boot --fstype ext4 --size=200 --ondisk=sda > part pv.6 --size=100 --grow --ondisk=sda > volgroup rootvg --pesize=32768 pv.6 > logvol swap --fstype=swap --name=swap --vgname=rootvg --size=2048 > logvol /var --fstype=ext4 --name=var --vgname=rootvg --size=2048 > logvol /var/tmp --fstype=ext4 --name=vartmp --vgname=rootvg --size=1024 > logvol /home --fstype=ext4 --name=home --vgname=rootvg --size=2048 > logvol /tmp --fstype=ext4 --name=tmp --vgname=rootvg --size=1024 > logvol /opt --fstype=ext4 --name=opt --vgname=rootvg --size=2048 > logvol / --fstype=ext4 --name=root --vgname=rootvg --size=4096 > > %packages > @core > @base > @java > lvm2 > gcc > gcc-c++ > net-snmp > sysstat > strace > ntp > screen > compat-libstdc++-33 > -NetworkManager* > -bluez* > -yum-updatesd > %end > > > Thanks, > > Andy > > _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list