[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] Windows XP & 2000 GUI Crashes due to Condor



Hi all,

I just updated to 6.7.2 the otehr day and just now i experienced a total crash on my machine . all the programs vanished without warning and windows explorer closed down..a few seconds later windows explorer started up again...My condor job is still computing away in the background.

Perhaps it is memory related. It seems to happen more often when condor job (Maya) is rendering in background and i am working at teh same time..usually reading email or web surfing..my computer only has 512 megs right now.

I'll try to create a memory eating php program to see if i can replicate the behaviour.

Condor master: Linux 6.7.2
Condor execute machine Windows XP COndor 6.7.2

JW




Jimmy Choi wrote:


Hello,

In 6.7.2, is anyone still seeing the problem where Condor would kill
processes that it doesn't own as part of the vacation process?


Apparently the problem is still there for WinXP (Win2K seems to be fine
but not sure). It'd be nice to get a confirmation about the problem, and
whether a complete fix is scheduled to come out soon.

Thanks
Jimmy


-----Original Message----- From: condor-users-bounces@xxxxxxxxxxx [mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of Jimmy Choi Sent: Wednesday, October 06, 2004 5:10 PM To: john@xxxxxxxxxx; Condor-Users Mail List Subject: RE: [Condor-users] Windows XP & 2000 GUI Crashes due to Condor

Hi John,

We are having the same problem here at Altera. The response that I
received from the Condor development team is that the problem should be
fixed in 6.6.6, but the fixes have not been fully merged back to the 6.7
series. I was told that 6.7.2 should contain most of them.


This is a very serious problem. Nobody would be willing to join their
machines to the Condor pool if they keep seeing their application
windows got killed one by one...

Jimmy

-----Original Message-----
From: condor-users-bounces@xxxxxxxxxxx
[mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of John Wheez
Sent: Tuesday, October 05, 2004 9:28 PM
To: Condor-Users Mail List
Subject: [Condor-users] Windows XP & 2000 GUI Crashes due to Condor

Hi,

I've noticed this alot in many versions of Condor. I use Condor linux 6.7.1 as my master & condor 6.7.1 windows XP on all client machines.

Sometimes when i am using a machine which is executing background condor

tasks, my windows GUI and all programs and windows are shut down.
Then the gui dissapears..a few seconds later the windows gui comes back but all programs that were running were killed. Also when this happens, condor_status and condor_q are unavailable on the linux condor master...its as if condor on the master crashes and some how it crashes all the client machines too. Awesome! but this is probably not the behaviour that is desired.



When this happens ALL condor processes on the Windows machines are killed so i need to manually restart them or reboot the machines. On the


master linux server after a minute or two condor_q produces and condor_status produce output. But incorrect output. Condor_status shows that the windows clients are busy computing but infact they are not because the applications were killed and condor is not even running on those machines anymore. Windows Task Manager shows the computer as 100% idle. Odd.

I suspect the Schedd deamon has something to do with this because when i

try to shut down the binaries linux waitys a while and says the processs

is "Disfunct"

To get condor running again, I have to shutdown all the processes on my

master and restart them on my master and all clients...funn fun.

My clients were not out of memory or diskspace either...no where close.

anyone else notice this behaviour?

JW


_______________________________________________ Condor-users mailing list Condor-users@xxxxxxxxxxx http://lists.cs.wisc.edu/mailman/listinfo/condor-users

_______________________________________________
Condor-users mailing list
Condor-users@xxxxxxxxxxx
http://lists.cs.wisc.edu/mailman/listinfo/condor-users