[molpro-user] Possible bug in precompiled binary

[Kliewer, Konrad C.]

Andy,

Thanks for your response.  It does make sense that that is what we are seeing since it only seems to happen when there are multiple molpro jobs running, in this case 5.  I have implemented the suggested fix and we will see how things run.

Thanks,
Konrad

Konrad Kliewer
Department of Chemistry
Purdue University
765-494-5235


-----Original Message-----
From: mayaj1 at Cardiff.ac.uk [mailto:mayaj1 at Cardiff.ac.uk] 
Sent: Thursday, April 04, 2013 4:43 AM
To: Kliewer, Konrad C.
Cc: molpro-user at molpro.net; Francisco, Joseph S; Mandy C Green
Subject: Re: [molpro-user] Possible bug in precompiled binary

Konrad,

This is not an error I've seen before, and I'm not even sure it's a bug in Molpro, but of course it could be that only Molpro is exposing the problem.

Putting 'INFO: task blocked for more than 120 sec' into google gives some results where the conclusion is either a slow shared file system or a problem with the disk. So I guess it would be worth verifying the disk.

This link:

http://blog.ronnyegner-consulting.de/2011/10/13/info-task-blocked-for-more-than-120-seconds/

seems relevant, the timing of the post seems to indicate it's relevant to your OS version, and it specifically mentions machines with large amounts of memory.

Best wishes,

Andy

On 20/03/13 18:49, Kliewer, Konrad C. wrote:
> Hi,
>
> I have a student running the 2012 precompiled molpro binary who keeps having his jobs crash.  The system he is using is a Dell R815 server with 4x12 core AMD processors with 512 GB of RAM.  We are running RHEL 5.5 on the system.  Following is the screen dump we are getting:
>
> INFO: task molpro.exe:29289 blocked for more than 120 sec Call Trace:
> [<ffffffff80016ee2>] generic_file_aio_read+0x36/0x3b 
> [<ffffffff80063c63>] __mutex_lock_slowpath+0x60/0x9b 
> [<ffffffff80063cad>] .test.lock.mutex+0xf/0x14
> [<ffffffff800baf21>]audit_syscall_entry+0x1a8/0x1d3
> [<ffffffff800e4b35>] generic_file_llseek_size+0x36/0x8e
> [<ffffffff80025bf3>] sys_lseek+0x40/0x60 [<ffffffff8005d23a>] 
> tracesys+0x71/0xdf [<ffffffff8005d29e>]tracesys+0xd5/0xdf
>
> Bunch of memory address that I can give if they are pertinent.
>
> Thanks in advance for any help given.
> ,
> Konrad
>
> Konrad Kliewer
> Department of Chemistry
> Purdue University
> 765-494-5235
>
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user
>