Hi,
Yes, I'd found that.
Digging a bit deeper I found occasional failures to stage files in from
the CE in the logs of some worker nodes.
Since this appears to be only sporadic I suspect I'm hitting the sshd
MaxStartups limit on the CE. I'd increased this in the past but had
forgotten to reapply the change when I reinstalled with SL4.
It's in now but I'll have to see if the errors disappear to know if it
was the root cause of the problem.
Thanks,
Chris.
> -----Original Message-----
> From: LHC Computer Grid - Rollout
> [mailto:[log in to unmask]] On Behalf Of Samuel
> Cadellin Skipsey
> Sent: 23 January 2008 15:32
> To: [log in to unmask]
> Subject: Re: [LCG-ROLLOUT] Pool account mapping difference
> between SL3 and SL4 CEs
>
> The GOC wiki directs me to the Unspecified Gridmanager Error page for
> details on Globus Error 94 (the page on Globus Error 94
> merely says "there
> was a problem with the batch system", which is helpful).
>
> http://goc.grid.sinica.edu.tw/gocwiki/Unspecified_gridmanager_error
>
|