Me and my brother are having strange problem with opensuse.
I run my computer (computer2) with 64 version of opensuse and my brother (computer1) with 32bit version.
It’s been about 15 days since both installs occured. With compiz fusion my computer slowed down after 4h just like computer was running out of memory but ‘free’ shows that there is half of ram unused and 1gb of swap unused. I dropped compiz fusion away and all was ok, except that about 7 days and it happened again, without compiz fusion.
I can not find anything in logs. Now it’s gone to 12h after reboot computer starts to slow as i described and after a while system dies. Ssh connection can not be made and apache seems to open socets but nothing else.
My brother have similar broblems. He haven’t noticed the slowing down but after a really short period (from 7 days to 12h that it’s now) computer just stops resbonding to anyting but unpluging the power cord.
We haven’t found anything from logs. No ioerrors no nothing.
While slowdown restarting services doesn’t work, and sometimes some of the services won’t get back up. xinetd, named, as for excample.
Hardware:
computer1:
dualxeon u1 server intel 1000 nics
computer2:
amd athlon 4400 x2
msi k8n neo4
2gb ram
500gb samsung sata disk
120gb samsung pata disk
nvidia 8500se
realteck nick
Have you checked the running processes when it slows?
I get some slowing but only when I’ve been using flash player plugin in a browser.
I have stopped using compiz for now and just use what effects there are in kde4 kwin. Much better, but I do miss compiz and emerald.
caf4926 wrote:
> Have you checked the running processes when it slows?
great question…
the brothers should (imo) save a snapshot of top when all is well, and
another when things have slowed down…
and/or install atop (which will, by default, take a snap every 10 minutes)…
and, i WONDER if the 12h means it is slow in late afternoon when the
room is (probably) the warmest…that is, MAYBE it is a heat problem
forcing the CPUs to throttle…
and, confirm that beagle is not the problem…and that you are not using
KDE4…
note: i did not say that it is not a bug!
–
DenverD (Linux Counter 282315)
A Texan in Denmark
Well we are open to all suggestions because we don’t have a clue what’s causing it.
Though I am sure that heating is not the problem.
We both have LVM partitioning system in use. I’m only throwing guesses what it could be. Beagle is worth of cheking, but it seams that there is no prosess eating all the memory or cpu time. I’m not sure if my brother even runs X11 if not needed, though I do and run kde3 with some kde4 applications
try giving beagle the 2 fingered salute
see if things improve
are either of you using reiserFS?
and, yes or no, have you installed all updates and patches?
(there was a recent kernel update to solve a problem with reiser and
beagle…but, that hound remains a problem yet)…
note: i’m still not saying you don’t have a bug…just trying to
eliminate the obvious/known reasons for v 11.0
slowdowns/crashes/unhappinesses/etc…
–
DenverD (Linux Counter 282315)
A Texan in Denmark
I use ext3 partitions with LVM, and as far as zypper we have all updates installed.
Disabled beagle, well see how it turns out
> Disabled beagle, well see how it turns out
be sure and come back with info on if that worked or not…
oh, if it does not work you might wanna install itop and check its log
files for the slow down culprit…
–
DenverD (Linux Counter 282315)
A Texan in Denmark
> oh, if it does not work you might wanna install itop and check its log
sorry, that should have been “atop” <http://www.atcomputing.nl/Tools/atop/>
–
DenverD (Linux Counter 282315)
A Texan in Denmark
Well it’s very unlikely in beryl, because my brother doesn’t have it installed, nor running
Uninstalling beryl didn’t do much good. Still suffering the same problem.
My brother was able to locate (atleast his problem, we might have different now) what’s the problem, cron daily runs updatedb and that fails. His server drops like dead bird.
I don’t (atleast for my knowledge) run cron daily and i have different permissions for updatedb, that might be because i got slowdown but not total crash
> My brother was able to locate (atleast his problem, we might have
> different now) what’s the problem, cron daily runs updatedb and that
> fails. His server drops like dead bird.
yes, updatedb is a cpu/disk and memory intensive program…you should
schedule it for an otherwise quiet period…like if you are gonna be
running a huge mail list between midnight and 1 AM (or play an online
game) that is NOT a good time to also cron updatedb…SCHEDULE it for
YOUR normal sleeping hours…or when the mail program is normally
quiet…etc
what is the error thrown by updatedb?
anyway, if he does not use ‘locate’ everyday he doesn’t need to daily
cron updatedb…especially during a period when it cripples the system…
> I don’t (atleast for my knowledge) run cron daily and i have different
> permissions for updatedb, that might be because i got slowdown but not
> total crash
ah yes permissions…why don’t you two make both of them look like the
one that does NOT crash…
and if he really doesn’t need locate ‘constantly’ it can be run from
the command line WHEN he needs the db updated…
–
DenverD (Linux Counter 282315)
A Texan in Denmark
The computer is idling while updatedb. And it does nearly everytime (note that not always) you run updatedb by hand
For me the excact problem is that computer slows down darn much, excample irssi gets so slow that its hard to use. Computer starts to drop from internet possibly because all is so slow that timeouts get done.
And the time that computer is so slow cpu is idling, there is plenty of memory for everything.
I might have finally cracked my problem!
It seams that nvidia driver was to reason weird slowdowns. I hope that new driver will fix it.
In my warn log was entry:
p, li { white-space: pre-wrap
kernel: NVRM: failed to unregister from the ACPI subsystem!
On just the same minute that computer went slowdown
> It seams that nvidia driver was to reason weird slowdowns. I hope that
> new driver will fix it.
if not, be SURE to post a bug to the nvidia folks…
> In my warn log was entry:
>
> Code:
> --------------------
> kernel: NVRM: failed to unregister from the ACPI subsystem!
> --------------------
but, i have to wonder why that didn’t turn up in any of the logs you checked
prior to writing in your first post: “We haven’t found anything from logs.”
–
DenverD (Linux Counter 282315)
A Texan in Denmark
Nope it didn’t help.
Also I can’t boot with acpi on, need to have it disabled, same with installed.
Incredibly I just didn’t noticed that error. Now that I look back those logs it is there. And it looks like causing the freezing problem.
I wonder if it is just opensuse kernel or if it is just that there is not good support for nforce4 chipset.
I also have this error
kernel: Clocksource tsc unstable
But as I have understand that if cpu can change frequency this shouldn’t be stable. But it apears always when freezing begins.
How could I configure different clock source or something?
/sys/devices/system/clocksource/clocksource0/available_clocksource shows that i have tsc and jiffies