Something funky is happing with the longer tasks

[zombie67]

I have been noticing this for a while now, but never had any evidence of it. I finally took some snapshots, and have something to show. I had 5 tasks running over 7 hours on an osx machine. I took the snapshot here:

https://files.me.com/zombie67/bggo21

This is the machine:

http://www.rnaworld.de/rnaworld/show_ho ... hostid=339

An hour or two later, I check back in on the machine, and they were all gone. So I look up the tasks for that machine on my account page. There are none that have run at least 7 hours. Not in pending, valid, invalid, or error. So where did they go? Are they one of the tasks in one of those lists, but showing with less time than 7 hours? Or missing all together?

[yoyo]

I used the first Wu in your PNG, which ends with *_1287247082_3* (this is the unique id). According to the server DB stats this wu is still on your host
http://www.rnaworld.de/rnaworld/workuni ... id=3543335
Can you check your boinc log what happend with this WU after computing?

yoyo

[mxplm]

Did you reboot your machine in that time?

[zombie67]

No reboot.

Here are the last couple of lines from the log:

Tue Oct 26 23:18:26 2010 RNA World Task cms_GA[e30-50MB_Lin64f]_Oryzias-latipes-(Japanese-medaka)_DG000012.lin.EMBL_RF00028_Intron_gpI_1287247082_3_4 exited with zero status but no 'finished' file
Tue Oct 26 23:18:26 2010 RNA World If this happens repeatedly you may need to reset the project.
Tue Oct 26 23:18:26 2010 RNA World Restarting task cms_GA[e30-50MB_Lin64f]_Oryzias-latipes-(Japanese-medaka)_DG000012.lin.EMBL_RF00028_Intron_gpI_1287247082_3_4 using cmsearch version 4

I see what is happening. It is starting over from scratch with 0 time on the clock. It has been 10 hours since this message, and it is now up to 10 hours of run time (again).

Edit: So now what? Is this a bug in the app?

[coturnix]

We are currently only providing i686 binaries for Mac OS X. Maybe the tasks just ran out of memory when they hit the 2 GB boundary? I mean with 32 bit there is only so much addresses available. But it's just a guess. @yoyo: what do you think about that? The one you pointed to was an Introl_gpI and we know they tend to consume quite much memory.

[zombie67]

We are currently only providing i686 binaries for Mac OS X. Maybe the tasks just ran out of memory when they hit the 2 GB boundary? I mean with 32 bit there is only so much addresses available. But it's just a guess. @yoyo: what do you think about that? The one you pointed to was an Introl_gpI and we know they tend to consume quite much memory.

For what it's worth:

1) Those 5 tasks are now over 16 hours of run time, since they restarted

2) This is an 8-core mac, with snow leopard, with 24gb of ram.

3) Current memory consumption is 7gb.

If it was a one-time thing, I would just shrug my shoulders and get on with life. But like I said in the first post, I have noticed this happening several times now.

[zombie67]

It happened again. All 5 tasks restarted from 0. They had over 30 hours of run time this time. Same error message.

Just for grins, I just now reset the project. Nothing to lose, as they had only a very little accumulated time. I got the same tasks back, which is normal. We'll see if that made any difference.

[zombie67]

It happened again. All 5 tasks restarted from 0. They had over 30 hours of run time this time. Same error message.

Just for grins, I just now reset the project. Nothing to lose, as they had only a very little accumulated time. I got the same tasks back, which is normal. We'll see if that made any difference.

Nope. Happened again 3 hours ago. I am aborting these 5 tasks.