Re: [HTCondor-users] HTCondor 8.0.6 tarball for SLES10?

[<Greg.Hitchen@xxxxxxxx>]

Thanks for all that info Daniel.

It would be "nice" if a version of condor 8.0.6 with those linker options
was available to test. I suspect it won't be high on the priority list
but would still be helpful to us (and maybe others?) if it's a simple task.

We will be going the route of replacing our current SLES10 VMs with
Ubuntu12 ones. Probably a good time to move from an out of date OS.

Cheers

Greg

-----Original Message-----
From: Daniel Forrest [mailto:dan.forrest@xxxxxxxxxxxxx] 
Sent: Monday, 24 March 2014 10:41 PM
To: HTCondor-Users Mail List
Cc: Hitchen, Greg (CESRE, Kensington)
Subject: Re: [HTCondor-users] HTCondor 8.0.6 tarball for SLES10?

Greg,

> OK, I've gone back to Condor 8.0.6 RH5 and the error is in fact not 
> mentioning libs but says:
> Floating point exception
> 
> If I "file condor_master" for 8.0.6 I get:
> ELF 64-bit LSB executable, x86-64, version 1 (SYSV), for GNU/Linux 
> 2.6.9, dynamically linked (uses shared libs), stripped
> 
> whereas for the working 7.6.7 I get:
> ELF 64-bit LSB executable, x86-64, version 1 (SYSV), for GNU/Linux 
> 2.4.0, dynamically linked (uses shared libs), stripped
> 
> which doesn't look that different?
> 
> As mentioned in my original email uname -r gives 2.6.16.60-0.91.1-smp
> 
> Any ideas why the FP exception?

This rings a bell...


I believe you are seeing the problem described here:

http://stackoverflow.com/questions/12570374/floating-point-exception-sigfpe-on-int-main-return0

Look at the first answer.

I verfied that the 8.0.6 condor_master does indeed have the GNU_HASH dynamic tag.


Also this link:

https://jira.mongodb.org/browse/SERVER-12427

Suggests that SLES 10 has this problem.


Perhaps you can get the Condor people to build a tarball with the "-Wl,--hash-style=both" linker option.

--
Dan