Stephen, surely a job runs as a particular uid/gid which is determined by things like gridmapfiles and pool accounts as the job passes through the CE. Changing these mappings after the job is submitted to the local queue should have no effect.
John
________________________________
From: Testbed Support for GridPP member institutes on behalf of Stephen Childs
Sent: Wed 22/04/2009 09:07
To: [log in to unmask]
Subject: Re: ATLAS STEP09 Site Request(s) + Hammercloud Tomorrow
Christopher J.Walker wrote:
>> If any site has a problem running tomorrow, please let me know ASAP.
>
> QM doesn't seem to be particularly fast at transferring data in and out,
> but I've no objection per se.
Two things:
1. Could someone explain to me how the transfers back to base work -- we
have around 600 jobs running and >2000 transferring. Who decides the rate
at which data gets transferred back? There shouldn't be a particular
limitation our end.
2. What happens if pool accounts etc. are reconfigured while jobs are
running? i.e. if someone was previously being mapped to a normal atlas
account and then starts getting mapped to a pilot pool account? Was also
thinking of changing to pool accounts for production role: similar issue
here (i.e. all users currently mapping to atlp will now be split up into
different pool mappings).
Stephen
--
Dr. Stephen Childs,
Research Fellow, EGEE Project, phone: +353-1-8961797
Computer Architecture Group, email: Stephen.Childs @ cs.tcd.ie
Trinity College Dublin, Ireland web: http://www.cs.tcd.ie/Stephen.Childs
--
Scanned by iCritical.
|