Ralph (& Rosetta) bring down BOINC Client

Message boards : RALPH@home bug list : Ralph (& Rosetta) bring down BOINC Client

To post messages, you must log in.

AuthorMessage
Profile m.mitch
Avatar

Send message
Joined: 12 May 06
Posts: 16
Credit: 154,608
RAC: 0
Message 3291 - Posted: 24 Jul 2007, 12:22:34 UTC
Last modified: 24 Jul 2007, 12:26:40 UTC

Both the RALPH and Rosetta projects are locking up the BOINC Clients and therefore all BOINC projects on my Win XP SP2 Home boxes. The BOINC Manger affected are 5.8.16, 5.10.7 and 5.10.13 on P4 3.2GHz Prescott, P4 540 3.2GHz, P4 2.6GHz Northwood (XP SP2 Pro), P3 1.2 GHz Tualatin but not that I've seen on a Ubuntu Linux P3 1.2 GHz Tualatin Server.

Unsuspending the current RALPH work unit causes the box to lock up for a moment, all the content in the BOINC Manger disappears and a fault report is sent to Microsoft. This is one of the problem RALPH work units: here

Although I don't know if the problem is caused by a conflict between projects, this is the other work unit that the same machine is working on here.

In the meantime, it isn't practical to run RALPH or Rosetta on unattended machines.


Click here to join the #1 Aussie Alliance on RALPH
ID: 3291 · Report as offensive    Reply Quote
Profile Conan
Avatar

Send message
Joined: 16 Feb 06
Posts: 364
Credit: 1,368,421
RAC: 0
Message 3293 - Posted: 31 Jul 2007, 11:14:37 UTC

> Hey Mike,
I am not sure of the problem but it seems to be Windows based.
I have an AMD 4800+ running Windows and it runs 24/7, projects it runs are QMC, Einstein, Seti, Rosetta and Ralph.
Ralph runs all the time and it is my best Rosetta machine, since last year I stopped using the graphics due to constant lock ups (with both Rosetta and Ralph) and hundreds of wasted hours of processing, and have had no lock ups since.

My other Linux machines don't have the problem due to graphics not being enabled in the first place.

The project staff claim that the problem was fixed but I gave up on the graphics and have not used them since.

So if you turn off the graphics your lock ups should go away (well mine did at least).

Catch you later.
ID: 3293 · Report as offensive    Reply Quote
Profile m.mitch
Avatar

Send message
Joined: 12 May 06
Posts: 16
Credit: 154,608
RAC: 0
Message 3294 - Posted: 31 Jul 2007, 15:15:37 UTC

G'day Conan,
Unfortunately I don't use the graphics either. So I'm still at a loss to work out what the problem is. It's difficult to trigger a fault, so I can't reproduce it at will and have no idea what the proximate cause is. Unlike Chess, it hates switching from one work unit to the next if it needs extra work. Then it does the same thing. Loses the BOINC Client.

That's VTU, RALPH, Chess (sometimes) and Rosetta that I can remember at the moment.


Click here to join the #1 Aussie Alliance on RALPH
ID: 3294 · Report as offensive    Reply Quote
Profile Conan
Avatar

Send message
Joined: 16 Feb 06
Posts: 364
Credit: 1,368,421
RAC: 0
Message 3295 - Posted: 1 Aug 2007, 12:39:45 UTC - in response to Message 3294.  

G'day Conan,
Unfortunately I don't use the graphics either. So I'm still at a loss to work out what the problem is. It's difficult to trigger a fault, so I can't reproduce it at will and have no idea what the proximate cause is. Unlike Chess, it hates switching from one work unit to the next if it needs extra work. Then it does the same thing. Loses the BOINC Client.

That's VTU, RALPH, Chess (sometimes) and Rosetta that I can remember at the moment.


I wonder if it could be memory related ? I believe that more than 512 MB would probably be a good thing.
I Doubt it could be graphic card related as you are not using them.
Are you using the latest or near latest Boinc client versions?
Switching between tasks does not seem to be a problem with my machines at all Windows or Linux. I am using 5.8.15 Windows and 5.8.16 Linux (I think, but deffinately the most recent prior to 5.10.x).
ID: 3295 · Report as offensive    Reply Quote
Profile m.mitch
Avatar

Send message
Joined: 12 May 06
Posts: 16
Credit: 154,608
RAC: 0
Message 3296 - Posted: 1 Aug 2007, 13:44:22 UTC - in response to Message 3295.  

I wonder if it could be memory related ? I believe that more than 512 MB would probably be a good thing.
I Doubt it could be graphic card related as you are not using them.
Are you using the latest or near latest Boinc client versions?
Switching between tasks does not seem to be a problem with my machines at all Windows or Linux. I am using 5.8.15 Windows and 5.8.16 Linux (I think, but definitely the most recent prior to 5.10.x).


I have 1GB on most machines and one with 2GB so there should be enough memory. There shouldn't be any faulty sticks, they've been running just fine for a while now and not causing any problems with other projects or programs.

I've tried the latest 5.8.x release and the two latest 5.10.x releases without success. The second last 5.8.x release, seemed okay but it wasn't loaded up like the current ones are. I'm running about 20 projects but one of the computers has always been running all the projects I'm attached too. It has me stumped.

May have to change them all over to Kubuntu, it's more stable anyway.


Click here to join the #1 Aussie Alliance on RALPH
ID: 3296 · Report as offensive    Reply Quote
Profile Conan
Avatar

Send message
Joined: 16 Feb 06
Posts: 364
Credit: 1,368,421
RAC: 0
Message 3297 - Posted: 2 Aug 2007, 13:01:38 UTC - in response to Message 3296.  

I wonder if it could be memory related ? I believe that more than 512 MB would probably be a good thing.
I Doubt it could be graphic card related as you are not using them.
Are you using the latest or near latest Boinc client versions?
Switching between tasks does not seem to be a problem with my machines at all Windows or Linux. I am using 5.8.15 Windows and 5.8.16 Linux (I think, but definitely the most recent prior to 5.10.x).


I have 1GB on most machines and one with 2GB so there should be enough memory. There shouldn't be any faulty sticks, they've been running just fine for a while now and not causing any problems with other projects or programs.

I've tried the latest 5.8.x release and the two latest 5.10.x releases without success. The second last 5.8.x release, seemed okay but it wasn't loaded up like the current ones are. I'm running about 20 projects but one of the computers has always been running all the projects I'm attached too. It has me stumped.

May have to change them all over to Kubuntu, it's more stable anyway.


Well Mike, good luck with trying to find the problem, I have no idea why this is occuring.
The Linux version I use is Fedora Core 3 and 6 (on AMD Opterons), this also is stable and runs fine.
My Windows machine on an ASUS M/B (bought for ability to over clock) has to run at default speeds due to the fact that as soon as you change the default setting (even just 1% to 3% speed wise) the Blue Screen of Death appears and I have to reboot. So Windows stays at default on the 4800+, so much for the O/C, took about a year to work this out.
ID: 3297 · Report as offensive    Reply Quote
Profile m.mitch
Avatar

Send message
Joined: 12 May 06
Posts: 16
Credit: 154,608
RAC: 0
Message 3298 - Posted: 3 Aug 2007, 5:10:17 UTC

Thanks Conan,
Think I'll just run RALPH & Rosetta on the Linux box from now on and save myself the headache.




Click here to join the #1 Aussie Alliance on RALPH
ID: 3298 · Report as offensive    Reply Quote
FluffyChicken

Send message
Joined: 17 Feb 06
Posts: 54
Credit: 710
RAC: 0
Message 3311 - Posted: 8 Aug 2007, 7:16:51 UTC
Last modified: 8 Aug 2007, 7:21:23 UTC

Hi, unfortunatly the error code jsut says you aborted the work unit.
But given most people are have no problems with Rosetta it must be your setup.
(Ralph I would at least expect some problems may happen as it is a testing project, though since noone else is reporting I doubt it is)

Given it's across multiple computers, what is common between them, some AV software, Firewall software etc..
OR as you've mentioned some project incompatability, Since QMC is running and that's also testing (RC at the moment ?)
You would need to see if Rosetta as the only running project (suspend the reset) causes the problem. QMC as the only running project casues the problem
Then both together.
You could probably speed problems up by altering the 'switch between projects' setting to a smaller value, say 15mins.

P.S. What are microsoft error messages saying casued the problem. You maybe able to find out easier in the Event Log.
ID: 3311 · Report as offensive    Reply Quote

Message boards : RALPH@home bug list : Ralph (& Rosetta) bring down BOINC Client



©2024 University of Washington
http://www.bakerlab.org