 |
 |
Actually, I now realise this could be edg-gridftp-ls making a mess of things. Files inside the DPM namespace have private UID/GIDs. What's important is the mapping in the DPNS user table: > use cns_db; > select * from Cns_userinfo; It would be easier to debug this is we were allowed access to port 5010 on the DPM box... g On 6 Jul 2006, at 01:47, Henry Nebrensky wrote: > On Wed, 5 Jul 2006, Graeme Stewart wrote: > >> On 5 Jul 2006, at 20:05, Dr D J Colling wrote: > ... >> Then: >> >> grid13:~$ edg-gridftp-ls --verbose gsiftp://dgc-grid-34.brunel.ac.uk: >> 2811/dpm/brunel.ac.uk/home/dteam >> ... >> -rw-rw-r-- 1 ops013 dteam 520 Jul 05 21:14 >> gs_test_20060705 >> ... >> >> Note the rather odd UID mapping to ops. I also see that with your >> test directory: >> >> grid13:~$ edg-gridftp-ls --verbose gsiftp://dgc-grid-34.brunel.ac.uk: >> 2811/dpm/brunel.ac.uk/home/cms >> drwxrwxr-x 0 ops012 cms 0 Jul 05 18:36 >> DJC_test >> >> Which indicates we're being rather aggressively mapped to the ops VO. > ... >> It's active, and seems to be ok (although 1.7GB to Brunel will take a >> while...) >> >> I suspect something might be up with CMS transfers in particular. Can >> you try a simple srmcp and check that that works. And are you mapped >> to some odd ops person? >> >> Perhaps Duncan or Henry could check the DPM mapping file. > > Duncan's away at the moment; I don't know if this is supposed > failover to > me... > > 1) At least some of the CMS Heartbeat transfers [*] worked earlier > "to"day > FWIW. Except RAL, I think... > > 2) Looked at gridmap-file on DPM head: > > "/C=UK/O=eScience/OU=Imperial/L=Physics/CN=david colling" .cms > "/C=UK/O=eScience/OU=Glasgow/L=Compserv/CN=graeme stewart" .dteam > > dpm/log > 07/05 17:45:16 3252,24 dpm_srv_inc_reqctr: DP092 - inc_reqctr > request by /C=UK/O=eScience/OU=Imperial/L=Physics/CN=david colling > (30012,1300) from dgc-grid-34.brunel.ac.uk > > dpns/log: > 07/05 17:45:16 3045,0 getidmap: Creating a new Virtual uid for / > C=UK/O=eScience/OU=Imperial/L=Physics/CN=david colling > 07/05 17:45:16 3045,0 Cns_srv_getidmap: NS092 - getidmap request > by /C=UK/O=eScience/OU=Imperial/L=Physics/CN=david colling > (30012,1300)... > > UID 30k+ is ops, gid 1300 is cms (as reported). That's my DPM > knowledge > exhausted - if someone can explain where getidmap is looking I'll > try to > have a look. > > 3) Elsewhere, Graeme Stewart wrote: > ... >> It's active, and seems to be ok (although 1.7GB to Brunel will take >> a while...) >> >> Actually, it timed out. FTS doesn't like transfers which take >> >~3600s... > > 1.7GB in 1 hour = 1.7MB in 3.6s = 13.6 Mb in 3.6 s = 3.8 Mb/s. > > So that ought to fit into our bandwidth reservation easily. Duncan > would > know if something else needs chasing. > > Thanks > > Henry > > * p.s. got bored at lunchtime and checked it. Most transfers > failed: same > problems/solutions as in May, should they ask. > > -- > Dr. Henry Nebrensky [log in to unmask] > http://people.brunel.ac.uk/~eesrjjn > "The opossum is a very sophisticated animal. > It doesn't even get up until 5 or 6 p.m." > -- Dr Graeme Stewart - http://wiki.gridpp.ac.uk/wiki/User:Graeme_stewart GridPP DM Wiki - http://wiki.gridpp.ac.uk/wiki/Data_Management