Boinc under FC5 linux keeps crashing with SIGSEGV

Questions and Answers : Unix/Linux : Boinc under FC5 linux keeps crashing with SIGSEGV
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Dougie Trickett

Send message
Joined: 17 May 99
Posts: 4
Credit: 1,700,938
RAC: 1
United Kingdom
Message 277661 - Posted: 7 Apr 2006, 22:43:24 UTC

Had this happening every workunit on FC4, so I wiped the drive and installed FC5 from scratch. Installed 5.2.13 i686-pc-linux-gnu and have successfully attached to seti and climateprediction.
Everthing goes OK for a while, then will not restart without :
SIGSEGV: segmentation violationStack trace (4 frames):
./boinc[0x80845b2]
[0xb24420]
./boinc[0x80784d9]
[0x5f6e306e]

I don't have any clue why this is happening and will likely remove boinc shortly if I can't get it going again.

Short rant: I have to say boinc is bitterly disappointing given the sublime ease "old" seti ran with. Sometimes change isn't good.
ID: 277661 · Report as offensive
Dotsch
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 2422
Credit: 919,393
RAC: 0
Germany
Message 277699 - Posted: 8 Apr 2006, 0:10:53 UTC
Last modified: 8 Apr 2006, 0:12:07 UTC

Installed 5.2.13 i686-pc-linux-gnu and have successfully attached to seti and climateprediction.
Everthing goes OK for a while, then will not restart without :
SIGSEGV: segmentation violationStack trace (4 frames):
./boinc[0x80845b2]
[0xb24420]
./boinc[0x80784d9]
[0x5f6e306e]

I don't have any clue why this is happening and will likely remove boinc shortly if I can't get it going again.

What is boinc doing excatly, when you get the SIGSEGV ? - Is it every time on the same stage, or is the SIGSEGV randomly ? - Which Application ist running, when you get the SIGSEGV.
Can you please post the messages from boinc, when you get the SIGSEGV.

Short rant: I have to say boinc is bitterly disappointing given the sublime ease "old" seti ran with. Sometimes change isn't good.

Sorry, but I feel this a little bit offense. There are a lot of differnt systems out there, which have no problems to run boinc and the different applications. Also there are a lot of different OSes (special Linux) wich differs heavly, so i could be that it will not run stable on every system. - Also a SIGSEGV, like your problem, is not easy to find. It could also a OS problem, and you can not point the failure per default to boinc.
The old seti classic had a lot of problems in the backend and missing a lot of needed features, and so its naturaly to eliminate the problems.
ID: 277699 · Report as offensive
Profile Dougie Trickett

Send message
Joined: 17 May 99
Posts: 4
Credit: 1,700,938
RAC: 1
United Kingdom
Message 277826 - Posted: 8 Apr 2006, 7:13:40 UTC - in response to Message 277699.  

I had this start to happen when I had Fedora Core 4 installed. It only started to happen about three to four weeks ago after running without any problems for months.
Other factors made me completely wipe my hard disk and put on a fresh installation of Fedora Core 5. I also installed boinc from a fresh download.

I attached to SETI and CLIMATEPREDICTION, both of which have ran for about a week until now. Every time I try to restart, it bombs out with this SIGSEGV error. OK, so it just may be a Fedora thing, I don't know, but whhy does it run for approximately a week before stopping very very dead?

Here's the complete xterm output when I try to run the client:

2006-04-08 08:10:52 [---] Starting BOINC client version 5.2.13 for i686-pc-linux-gnu
2006-04-08 08:10:52 [---] libcurl/7.14.0 OpenSSL/0.9.8 zlib/1.2.3
2006-04-08 08:10:52 [---] Data directory: /home/dougie/BOINC
2006-04-08 08:10:52 [---] Processor: 1 AuthenticAMD AMD Athlon(tm) XP 2400+
2006-04-08 08:10:52 [---] Memory: 1.24 GB physical, 1.94 GB virtual
2006-04-08 08:10:52 [---] Disk: 34.13 GB total, 26.06 GB free
2006-04-08 08:10:52 [SETI@home] Computer ID: 2308767; location: home; project prefs: default
2006-04-08 08:10:52 [climateprediction.net] Computer ID: 428560; location: ; project prefs: default
2006-04-08 08:10:52 [---] General prefs: from SETI@home (last modified 2005-11-22 07:52:49)
2006-04-08 08:10:52 [---] General prefs: no separate prefs for home; using your defaults
2006-04-08 08:10:52 [---] Remote control not allowed; using loopback address
SIGSEGV: segmentation violationStack trace (4 frames):
./boinc[0x80845b2]
[0xb24420]
./boinc[0x80784d9]
[0x5f6e306e]

Exiting...
ID: 277826 · Report as offensive
Profile Dougie Trickett

Send message
Joined: 17 May 99
Posts: 4
Credit: 1,700,938
RAC: 1
United Kingdom
Message 277849 - Posted: 8 Apr 2006, 10:15:20 UTC - in response to Message 277699.  

OK, fixed.
After some searching, it appears the climate prediction project is the culprit. Both Sulphur cycle and BBC versions seem broken for Linux only and not an MS Windows problem.
Shame really, as I wasn't participating for long, but it looks like I will not be running for them under Linux at least.
ID: 277849 · Report as offensive
Dotsch
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 2422
Credit: 919,393
RAC: 0
Germany
Message 277852 - Posted: 8 Apr 2006, 10:30:13 UTC - in response to Message 277849.  

OK, fixed.
After some searching, it appears the climate prediction project is the culprit. Both Sulphur cycle and BBC versions seem broken for Linux only and not an MS Windows problem.
Shame really, as I wasn't participating for long, but it looks like I will not be running for them under Linux at least.

Oh. Do you have some links, which describes the CPDN and BBC bugs. - I did not known about this, because I am not computing this both projects. - Are this application problems fixable ?
The output you have posted looked, that it is a problem before starting the appliacations, and looked like a boinc or linux issue....

ID: 277852 · Report as offensive
Profile Dougie Trickett

Send message
Joined: 17 May 99
Posts: 4
Credit: 1,700,938
RAC: 1
United Kingdom
Message 277913 - Posted: 8 Apr 2006, 12:58:25 UTC - in response to Message 277852.  

Oh. Do you have some links, which describes the CPDN and BBC bugs. - I did not known about this, because I am not computing this both projects. - Are this application problems fixable ?
The output you have posted looked, that it is a problem before starting the appliacations, and looked like a boinc or linux issue....

As soon as I detatched from that project, boinc immediately started to work again.
I "googled" for "segmentation violationStack" and found quite a lot of people who were in the same boat as I was.

http://www.climateprediction.net/board/viewtopic.php?t=3659&sid=e9d81af980c31741f0fd415b2dc20651
Has the story.
Whether they are fixable, I don't know. I may check back when I have more time.
ID: 277913 · Report as offensive
Dotsch
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 2422
Credit: 919,393
RAC: 0
Germany
Message 278560 - Posted: 9 Apr 2006, 14:42:21 UTC - in response to Message 277913.  

http://www.climateprediction.net/board/viewtopic.php?t=3659&sid=e9d81af980c31741f0fd415b2dc20651

Looks like a linux or application problem.
Here are something, which you can also try to get BBC or CPDN to work :
http://bbc.cpdn.org/forum_thread.php?id=522#2022. - Also you could try to to set the max data, max text and max stack size kernel parameters higher.
ID: 278560 · Report as offensive
regisr
Volunteer tester

Send message
Joined: 22 May 04
Posts: 7
Credit: 10,668
RAC: 0
France
Message 282007 - Posted: 14 Apr 2006, 20:09:14 UTC - in response to Message 277826.  

OK, so it just may be a Fedora thing, I don't know, but whhy does it run for approximately a week before stopping very very dead?


The same problem occurs with FreeBSD 6 (production release) on my computer and with FreeBSD 7 (developpement) for another FreeBSD user.
I have only SETI client. This was (at least for my use) after I upgraded from 4.x to 5.2.15.
I find that the problems seems to be in the function curl_global_init
(which is not thread-safe according the man page) ...
call is in the source file net_xfer_curl.C.
(not with debugger with printing messages ;-)
And libcurl was upgraded from 7.15.1 to 7.15.3 at the same time
I asked on the ports mailing list for FreeBSD but they suggested to post on Seti/Boinc forums.


ID: 282007 · Report as offensive
Robert Huff

Send message
Joined: 29 Apr 99
Posts: 1
Credit: 1,847,660
RAC: 0
United States
Message 282058 - Posted: 14 Apr 2006, 21:57:15 UTC - in response to Message 282007.  


The same problem occurs with FreeBSD 6 (production release) on my computer and with FreeBSD 7 (developpement) for another FreeBSD user.



Same here. Under

FreeBSD 7.0-CURRENT #0: Mon Mar 13 09:23:39 EST 20

running

client-5.2.15
setiathome-4.18

I get what is almost certainly the idenitcal SEGV. (I have a core-dump with symbols and am willing to do backtraces if asked.)



ID: 282058 · Report as offensive
Dotsch
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 2422
Credit: 919,393
RAC: 0
Germany
Message 282110 - Posted: 14 Apr 2006, 23:04:50 UTC - in response to Message 282007.  
Last modified: 14 Apr 2006, 23:07:23 UTC

The same problem occurs with FreeBSD 6 (production release) on my computer and with FreeBSD 7 (developpement) for another FreeBSD user.
I have only SETI client. This was (at least for my use) after I upgraded from 4.x to 5.2.15.
I find that the problems seems to be in the function curl_global_init
(which is not thread-safe according the man page) ...
call is in the source file net_xfer_curl.C.
(not with debugger with printing messages ;-)
And libcurl was upgraded from 7.15.1 to 7.15.3 at the same time
I asked on the ports mailing list for FreeBSD but they suggested to post on Seti/Boinc forums.

It looks like a FreeBSD and boinc related problem. Specialy, looks like a problem with curl.
There is also a other thead (http://setiathome.berkeley.edu/forum_thread.php?id=29532) with simliar Problems under FreeBSD. Dougie had some CPDN app related problems.

But Stefan Urbat has a running and public donwloadable boinc client (http://www.lb.shuttle.de/apastron/boincDown.shtml#freebsd) bundled with the seti ehanced (=beta project) application. Could you please try his binaries out. It worked for a lot of users. - You can unpack it and take the boinc_client binay for furhter tests.

ID: 282110 · Report as offensive
Profile Crunch3r
Volunteer tester
Avatar

Send message
Joined: 15 Apr 99
Posts: 1546
Credit: 3,438,823
RAC: 0
Germany
Message 282112 - Posted: 14 Apr 2006, 23:06:11 UTC - in response to Message 282058.  


The same problem occurs with FreeBSD 6 (production release) on my computer and with FreeBSD 7 (developpement) for another FreeBSD user.



Same here. Under

FreeBSD 7.0-CURRENT #0: Mon Mar 13 09:23:39 EST 20

running

client-5.2.15
setiathome-4.18

I get what is almost certainly the idenitcal SEGV. (I have a core-dump with symbols and am willing to do backtraces if asked.)




If the default app. isn't working for you, then you could try my optimized one.
(see the link in my sig.)

It requires at least an cpu capable of SSE instruction set and it's at least 3 times faster than the app. you're getting from "ports" or berkeley.






Join BOINC United now!
ID: 282112 · Report as offensive
regisr
Volunteer tester

Send message
Joined: 22 May 04
Posts: 7
Credit: 10,668
RAC: 0
France
Message 282425 - Posted: 15 Apr 2006, 17:08:00 UTC - in response to Message 282110.  
Last modified: 15 Apr 2006, 17:12:41 UTC


But Stefan Urbat has a running and public donwloadable boinc client (http://www.lb.shuttle.de/apastron/boincDown.shtml#freebsd) bundled with the seti ehanced (=beta project) application. Could you please try his binaries out. It worked for a lot of users. - You can unpack it and take the boinc_client binay for furhter tests.


It is working... At least it is downloading the WU ;-)
Robert, did you installed it?

For install it I put the files (without backing up the previous ;-) :
projects/setiweb.ssl.berkeley.edu_beta/setiathome-5.5.i386-unknown-freebsd
in /var/db/boinc/projects/setiathome.berkeley.edu as setiathome

projects/setiweb.ssl.berkeley.edu_beta/app_info.xml in /var/db/boinc/projects/setiathome.berkeley.edu

boinc5.2.14-freebsd-i386-mmx in /usr/local/bin as boinc_client

Be carefull: the archive don't have a top directory. I desarchived on a new directory.

Pav wrote me that the port boinc-client-5.4.3 should be avalaible soon.

ID: 282425 · Report as offensive
Dotsch
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 2422
Credit: 919,393
RAC: 0
Germany
Message 282435 - Posted: 15 Apr 2006, 17:48:30 UTC - in response to Message 282425.  
Last modified: 15 Apr 2006, 17:54:53 UTC

For install it I put the files (without backing up the previous ;-) :
projects/setiweb.ssl.berkeley.edu_beta/setiathome-5.5.i386-unknown-freebsd
in /var/db/boinc/projects/setiathome.berkeley.edu as setiathome

projects/setiweb.ssl.berkeley.edu_beta/app_info.xml in /var/db/boinc/projects/setiathome.berkeley.edu

No. The setiathome-5.5 binary and the app_info.xml are for the beta project and not for the production seti !
You must use your old setiathome-4.x binary and the old app_info.xml for the production project.


ID: 282435 · Report as offensive
regisr
Volunteer tester

Send message
Joined: 22 May 04
Posts: 7
Credit: 10,668
RAC: 0
France
Message 282581 - Posted: 15 Apr 2006, 22:28:43 UTC - in response to Message 282435.  


No. The setiathome-5.5 binary and the app_info.xml are for the beta project and not for the production seti !
You must use your old setiathome-4.x binary and the old app_info.xml for the production project.


You are right! I apologize for this error.

I have do the following:
Copy only the boinc_client to /usr/local/bin and keep the setiathome 4.18.
Don't desinstall the port: other files are not included in the tar file.
The setiweb.ssl.berkeley.edu_beta folder should be copied as it in the projects folder for the beta project but I can't register for this project. (it is a new account).

ID: 282581 · Report as offensive

Questions and Answers : Unix/Linux : Boinc under FC5 linux keeps crashing with SIGSEGV


 
©2025 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.