On 02/02/11 18:50, Daniela Bauer wrote:
[ ... ]
> The common factor is RAL lcgwms03 - I've reassigned my ticket to RAL.
My ticket (and these are local users) is https://gus.fzk.de/ws/ticket_info.php?ticket=66928 and it has the remarkable point that the failures via ScotGrid WMSes are related to our CE02 but not CE01 (two hopefully soon to be replaced lcg-CEs):
> [ ... ] In Durham all jobs submitted via RAL failed due to the proxy expiring, [ ... ]
> While most jobs to ce01 via Glasgow succeeded all the jobs to ce02 failed [ ... ]
I have been perplexed by some differences in behaviour between them in the past (e.g. the Globus gatekeeper crashes regularly on CE02 but not CE01), and I guess they had been setup to be essentially identical; even if they seem to get quite different loads from the ScotGrid WMSes.
I have just done, as a first check, a 'diff' between the '/etc' and '/opt/*/etc' directories and they seem virtually identical apart from the host name.
Please let me know suggestions about where to look to what makes CE01/CE02 behave differently.
|