On 09/08/12 15:02, Alessandra Forti wrote:
> Hi,
>
> I'm trying to redress some space imbalances, i.e. one server too full.
> So I'm running dpm-drain
>
> dpm-drain --server server.blah.blah.blah --size 2T
>
> This works fine if left going but often we have entries like these.It
> seems the space token was deleted and perhaps recreated. Are these
> files still accessible by atlas at any rate? Can they be easily found
> and deleted? Dear I could grep the dpm-drain output but maybe there is
> a better way.
>
I think they are potentially accessible - but it seems very unlikely
that they would be.
To find them in the DB I think it is just
select Cns_file_replica from Cns_file_replica where
Cns_file_replica.setname='9/fc7cb06-8644-45dc-bb97-e7398c5b14d7/'
> /08/09 14:50:13 21524,0: File:
> /dpm/tier2.hep.manchester.ac.uk/home/atlas/atlasgroupdisk/SAM/testfile-put-ATLASGROUPDISK-1290698761-0bd0e57247d8.txt
> 08/09 14:50:13 21524,0: pfn:
> se05.tier2.hep.manchester.ac.uk:/raid/atlas/2010-11-25/testfile-put-ATLASGROUPDISK-1290698761-0bd0e57247d8.txt.8528773.0
> (of 1)
> 08/09 14:50:13 21524,0: replica type: primary
> 08/09 14:50:13 21524,0: file type: permanent08/09 14:50:13
> 21524,0: (does not expire)
> 08/09 14:50:13 21524,0: size: 20
> 08/09 14:50:13 21524,0: space:
> 9fc7cb06-8644-45dc-bb97-e7398c5b14d7dpm_getspacemd: Invalid/unknown
> space token
> 08/09 14:50:13 21524,0: (invalid space)
> 08/09 14:50:13 21524,0: replicating...
> dpm_getstatus_putreq: Failed for all SURLs
> 08/09 14:50:15 21524,0: failed
> 08/09 14:50:15 21524,0: dpm_replicatex
> se05.tier2.hep.manchester.ac.uk:/raid/atlas/2010-11-25/testfile-put-ATLASGROUPDISK-1290698761-0bd0e57247d8.txt.8528773.0:
> No space left on device
> /
>
> We also have entries like these were the connection gets closed by the
> remote host but the only remote host that doesn't accept connections
> because I set the weight to 0 is the host I'm trying to drain. Isn't
> that excluded by the drain list of possible targets?
>
> /08/09 14:51:37 21524,0: File:
> /dpm/tier2.hep.manchester.ac.uk/home/atlas/atlasdatadisk/mc10_7TeV/AOD/e574_s933_s946_r1652_r1700/mc10_7TeV.108319.PythiaDrellYan_mumu.merge.AOD.e574_s933_s946_r1652_r1700_tid192277_00/AOD.192277._000015.pool.root.1__DQ2-1295525516
> 08/09 14:51:37 21524,0: pfn:
> se05.tier2.hep.manchester.ac.uk:/raid/atlas/2011-01-20/AOD.192277._000015.pool.root.1__DQ2-1295525516.10431839.0
> (of 1)
> 08/09 14:51:37 21524,0: replica type: primary
> 08/09 14:51:37 21524,0: file type: permanent08/09 14:51:37
> 21524,0: (does not expire)
> 08/09 14:51:37 21524,0: size: 697.85M
> 08/09 14:51:37 21524,0: space:
> 685ca746-a4aa-4a0e-80a5-b52cc3acdf1c08/09 14:51:37 21524,0:
> (ATLASDATADISK)
> 08/09 14:51:37 21524,0: replicating...
> 08/09 14:51:38 21524,0: failed
> 08/09 14:51:38 21524,0: dpm_replicatex
> se05.tier2.hep.manchester.ac.uk:/raid/atlas/2011-01-20/AOD.192277._000015.pool.root.1__DQ2-1295525516.10431839.0:
> Connection closed by remote en/
I didn't quite follow the issue here. Are you saying it is trying to
drain onto a filesystem even though the weight is 0 . (that would be
annoying if it does that).
wahid
> thanks
>
> cheers
> alessandra
> --
> Facts aren't facts if they come from the wrong people. (Paul Krugman)
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
|