Welcome to MilkyWay@home

Milky Way processes going zombie?

Message boards : Number crunching : Milky Way processes going zombie?
Message board moderation

To post messages, you must log in.

AuthorMessage
Donald Qualls

Send message
Joined: 13 Apr 11
Posts: 33
Credit: 29,536,820
RAC: 5,393
Message 59557 - Posted: 9 Aug 2013, 1:34:55 UTC

Several times recently I've noticed I don't have the long yellow bars ("nice" priority tasks on each CPU) in my system monitor widget, and on checking I find the Milky Way tasks have stalled in some fashion, getting reported as "zombie" -- and I can't individually stop/kill those tasks, I have to kill the entire BOINC client, and then restarting BOINC manager fails to restart the client or its tasks (hangs at "connecting to local host"); I have to do a full system restart to get BOINC back in operation (logout/login won't fix it; BOINC runs as a system task and loads before login).

For whatever it's worth, I'm running BOINC 6.10.58 in MEPIS 11 Linux 64-bit (can't find a newer version in the repos recommended for MEPIS 11/Debian Squeeze, even Testing, can't seem to get alioth to show me packages, and haven't ever gotten the direct download Berkeley installer to work), on a 2.5 GHz Intel E5200 (dual core), 4 GiB RAM, nVidia GT520 w/ 1 GiB (which doesn't get me GPU tasks on Milky Way), lots of hard disk space free, and haven't installed anything new recently.
ID: 59557 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3319
Credit: 520,373,633
RAC: 23,342
Message 59559 - Posted: 9 Aug 2013, 10:55:43 UTC - in response to Message 59557.  

Several times recently I've noticed I don't have the long yellow bars ("nice" priority tasks on each CPU) in my system monitor widget, and on checking I find the Milky Way tasks have stalled in some fashion, getting reported as "zombie" -- and I can't individually stop/kill those tasks, I have to kill the entire BOINC client, and then restarting BOINC manager fails to restart the client or its tasks (hangs at "connecting to local host"); I have to do a full system restart to get BOINC back in operation (logout/login won't fix it; BOINC runs as a system task and loads before login).

For whatever it's worth, I'm running BOINC 6.10.58 in MEPIS 11 Linux 64-bit (can't find a newer version in the repos recommended for MEPIS 11/Debian Squeeze, even Testing, can't seem to get alioth to show me packages, and haven't ever gotten the direct download Berkeley installer to work), on a 2.5 GHz Intel E5200 (dual core), 4 GiB RAM, nVidia GT520 w/ 1 GiB (which doesn't get me GPU tasks on Milky Way), lots of hard disk space free, and haven't installed anything new recently.


First I am NOT a Linux guy, but in Windows if you suspend the project and then after a slow 5 count resume it again, the stuck units should pick back up and get crunching again. I have not experienced it here at MW but at other projects this works just fine.
ID: 59559 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : Milky Way processes going zombie?

©2024 Astroinformatics Group