[K12OSN] 8 Minutes to Open Open Office 2
R. Scott Belford
scott at hosef.org
Fri Oct 7 19:34:27 UTC 2005
Edward Holcroft wrote:
> I was not able to replicate this scenario on another server, this time
> a single CPU machine. However - I now note that on the first server, I
> get the output referred to above when I run javaldx. On the server
> which still runs OOo slowly, I get no output at all when I run javaldx.
> Could this be the problem?
>
> ed
Aloha Amigos
My apologies for not following up more timely on this issue. A faithful
volunteer, Ron, went by one of the trouble schools last week and came up
with these metrics. I have not included all of his measurements, but
easily can upon request.
In summary, the lab runs great, except for OO. While some are waiting
for it to open, others are actively using it. Meanwhile, the
application menus are snappy, other apps readily open, etc. At no time
in top is the cpu or ram totally utilized. In fact, three instances of
firefox had more cpu utilization than all OO instances.
(scott)
*mii-tool* (ha! you guys were so right. the switch light does lie)
[root at kuhio-hosef ~]# mii-tool
eth0: negotiated 100baseTx-FD, link ok
eth1: negotiated 10baseT-FD flow-control, link ok
*javaldx* (this lab was an upgrade, not a clean install)
[root at kuhio-hosef ~]# /usr/lib/openoffice.org2.0/program/javaldx
/usr/java/j2re1.4.2_02/lib/i386/client:/usr/java/j2re1.4.2_02/lib/i386/native_threads:/usr/java/j2re1.4.2_02/lib/i386
*Observational* *Narrative*
I was onsite at Kuhio school with Debbie yesterday during and after one
of her large classes that was experiencing the extreme slowness when
opening OpenOffice on the K12LTSP server. I'm attaching logs taken
during this time. /var/log/messages was tailed the entire time to
tail-messages.log. Shell script logit was run periodically to capture
the output of "ps auxwwf", "top -b -n1" and "free". "prelink" was run,
per the suggestion of one of the helpful folk on the k12osn mailing
list. See Scott's thread "[K12OSN] 8 Minutes to Open Open Office 2" in
the k12osn mailing list archives for Sep and Oct at
https://www.redhat.com/archives/k12osn/ .
Other files are the output of "df", "ifconfig", "rpm -qa | grep -i" for
k12ltsp, office, kernel and fedora.
The switch seemed to be working fine as far as i could see all port
activity lights flashing green, no amber or red lights, the switch is a
Amer.com SR48G2i connected via RS232 to the server (for console
management?).
DNS is working fine from the server, able to "dig outside.host.names"
and "dig -x outside.ip.addresses", all client workstations are in
the /etc/hosts file.
Each client logon is accompanied by an error dialog saying "No volume
control elements and/or devices found". Each logoff is accompanied by a
similar message but it disappears too quickly for me to get the full
text. Debbie has a fix for this problem from Ted but hasn't had time to
apply it yet.
The client logons proceed quickly and everything is pretty responsive
until OOwriter is invoked. The OpenOffice splash screen appears quickly
and the progress bar goes immediately to the exact halfway point and
then stalls. While OO is stalled, other apps on the same workstation and
OO itself on other workstations are all very responsive. A fair number
of times OO crashed before completing startup. I've included the
"oo.crash.log" text from one of these.
Another anomaly noted was that three of the users had strange writer
behavior in that when they typed one or more "space" characters in the
active document, the cursor would not move until a non-space character
was typed, then the accumulated space characters would become visible. I
didn't test to see if this problem follows the user or the client
workstation.
Another point, 3 users were running mozilla (see ps-auxwwf.log and
top-b-n1.log), consuming a lot of cpu but I don't think this had
anything to do with the slowness opening OOwriter.
*Observational* *Summary*
Some times of note:
13:33:03 started tailing /var/log/messages with
"tail -f /var/log/messages | tee tail-messages.log"
13:35:15 last user from the 12:35 - 13:20 class logs out
13:44:37 first users from the 13:20 - 14:05 start to login
13:50 lots of kids are waiting for OOwriter to complete startup
others are happily working on their assignments.
13:51:54 kuhio-hosef kernel: 3w-xxxx: scsi2: AEN: WARNING:
SMART threshold exceeded: Port #1. [same message seen in
dmesg.log - a drive may be about to fail]
14:22:40 last user exits
*Memory* *Usage*
Fri Sep 30 13:22:18 HST 2005
total used free shared buffers cached
Mem: 2075804 2025296 50508 0 18732 211728
-/+ buffers/cache: 1794836 280968
Swap: 1052152 308952 743200
Fri Sep 30 13:22:55 HST 2005
total used free shared buffers cached
Mem: 2075804 2023436 52368 0 19396 211864
-/+ buffers/cache: 1792176 283628
Swap: 1052152 308952 743200
Fri Sep 30 13:24:48 HST 2005
total used free shared buffers cached
Mem: 2075804 2023436 52368 0 21980 213312
-/+ buffers/cache: 1788144 287660
Swap: 1052152 303160 748992
Fri Sep 30 13:25:35 HST 2005
total used free shared buffers cached
Mem: 2075804 2024908 50896 0 20808 196892
-/+ buffers/cache: 1807208 268596
Swap: 1052152 310856 741296
Fri Sep 30 13:36:41 HST 2005
total used free shared buffers cached
Mem: 2075804 596384 1479420 0 30580 251432
-/+ buffers/cache: 314372 1761432
Swap: 1052152 47156 1004996
Fri Sep 30 13:50:40 HST 2005
total used free shared buffers cached
Mem: 2075804 1918612 157192 0 49604 355488
-/+ buffers/cache: 1513520 562284
Swap: 1052152 37960 1014192
Fri Sep 30 13:51:31 HST 2005
total used free shared buffers cached
Mem: 2075804 2025128 50676 0 37124 298360
-/+ buffers/cache: 1689644 386160
Swap: 1052152 47056 1005096
Fri Sep 30 14:10:54 HST 2005
total used free shared buffers cached
Mem: 2075804 615404 1460400 0 26344 260488
-/+ buffers/cache: 328572 1747232
Swap: 1052152 80992 971160
Fri Sep 30 14:11:12 HST 2005
total used free shared buffers cached
Mem: 2075804 595796 1480008 0 26408 260568
-/+ buffers/cache: 308820 1766984
Swap: 1052152 80684 971468
*Open* *Office* *Log*
0x27bb96: /usr/lib/openoffice.org2.0/program/libuno_sal.so.3 + 0x1db96
0x27c3e4: /usr/lib/openoffice.org2.0/program/libuno_sal.so.3 + 0x1e3e4
0x879420: + 0x420 (__kernel_sigreturn + 0x0)
0x24b77e: /lib/ld-linux.so.2 + 0x977e
0x24f6d8: /lib/ld-linux.so.2 + 0xd6d8
0x254d30: /lib/ld-linux.so.2 + 0x12d30
0x5dc14b8: /usr/lib/libcups.so.2 + 0x54b8
0x5dc1cff: /usr/lib/libcups.so.2 + 0x5cff (cupsGetDests2 + 0x54)
0x5dc1fef: /usr/lib/libcups.so.2 + 0x5fef (cupsGetDests + 0x48)
0x8ca8b4: /usr/lib/openoffice.org2.0/program/libpsp680li.so + 0x508b4
0x8c876f: /usr/lib/openoffice.org2.0/program/libpsp680li.so + 0x4e76f
(psp::CUPSManager::runDests() + 0x21)
0x8c87c5: /usr/lib/openoffice.org2.0/program/libpsp680li.so + 0x4e7c5
(psp::CUPSManager::runDestThread(void*) + 0x17)
0x275339: /usr/lib/openoffice.org2.0/program/libuno_sal.so.3 + 0x17339
0x4b9b80: /lib/libpthread.so.0 + 0x5b80
0xe3edee: /lib/libc.so.6 + 0xcadee (__clone + 0x5e)
Hope this helps some figure out what's happening.
-- Ron Fox
More information about the K12OSN
mailing list