ogr WUs crash immediately after starting

[MLx]

Hi, I'm running BOINC on Linux (amd64) and any ogr WU I get crashes almost as soon as it starts (within 10 seconds).

<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
process got signal 11
</message>
<stderr_txt>
wrapper: starting
running dnetc
filename: ../../projects/www.rechenkraft.net_yoyo/ogr_110516150129_45_1_1

</stderr_txt>
]]>

Admins: See this WU or any other ogr WU on that host
I've completed ogr WUs on this machine before, but that was admittedly some 2-4 months ago. Has something changed since?

[yoyo]

Hello,
in the last time the dnetc version was updated.

Please make a project reset in your Boincmanager and try again.

If this doesn't help, on the command line cd into projects/www.rechenkraft.net_yoyo folder and run a
ldd dnetc519b-linux-amd64-elf
and post the output here.

thx
yoyo

[MLx]

Project reset didn't help. This is the result - the dnetc executable runs (and eats CPU), even though the WU is "finished" with error. As new WUs arrive and are started, the computer is littered with orphaned dnetc processes:

ogr_fail.png (131.06 KiB) 5450 mal betrachtet

The ldd command reports this:

Code: Alles auswählen

# ldd dnetc519b-linux-amd64-elf 
        linux-vdso.so.1 =>  (0x00007fffa21ff000)
        libesets_pac.so => /usr/lib/libesets_pac.so (0x00007f915b5c6000)
        librt.so.1 => /lib/librt.so.1 (0x00007f915b3be000)
        libc.so.6 => /lib/libc.so.6 (0x00007f915b03b000)
        libpthread.so.0 => /lib/libpthread.so.0 (0x00007f915ae1e000)
        libdl.so.2 => /lib/libdl.so.2 (0x00007f915ac1a000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f915b740000)

I'll try to detach and re-attach the project, I had it do the trick for another project where a reset did not fix things.

EDIT: Nope, detach-reattach doesn't help either.

[MLx]

BUMP!
Any ideas?

[yoyo]

No idea.
Which hostID is it?

Can you start
dnetc519b-linux-amd64-elf
on the command line?

yoyo

[MLx]

Host ID: 44474
Running dnetc519... from the commandline asks for some configuration:

root@esprimo:/var/lib/boinc-client/projects/www.rechenkraft.net_yoyo# sudo -u boinc ./dnetc519b-linux-amd64-elf
distributed.net client configuration:
--------------------------------------------------------------------------
1) General Client Options
2) Buffer and Buffer Update Options
3) Performance related options
4) Logging Options

9) Discard settings and exit
0) Save settings and exit

Note: You have not yet provided a distributed.net ID.
Please go to the 'General Client Options' and set it.

Choice -->

[MLx]

I got it! The app works now that I've uninstalled NOD32 antivirus. Not sure who's at fault here - I only had a trial version that expired recently so I won't be installing it again any time soon, but yoyo devs and users might want to know.

[yoyo]

Thanks for the note!
yoyo

[Crystal Pellet]

It looks like most of the current ogr-tasks end in a compute error:

process exited with code 28 (0x1c, -228)

[yoyo]

Do you have some example workunits?
yoyo

[yoyo]

As I can see, the errors are all from an PowerPC with Darwin 8.x.
yoyo

[Crystal Pellet]

Yoyo wrote:

As I can see, the errors are all from an PowerPC with Darwin 8.x.

You're right about that. But it isn't an (1) PC, but it looks like all MACs seem to have this problem.

http://www.rechenkraft.net/yoyo/workuni ... id=9408289