Hi,
I noticed some strange SFT behaviour this morning - simplest lcg-cr is stuck for 1 hour, producing CPU load.
I tested lcg-cr from my UI agains my BDII and it didn't work - same problem - it downloads some data from the BDII and
then hangs forever.
I changed LCG_GFAL_INFOSYS to lcg-bdii.cern.ch:2170 everywhere, since this appeared to work from the UI.
Then I tested several other BDIIs, which are at 2_6_0 (or 2_5_0), and I noticed that they exhibit the same problem.
Right now I can enjoy watching
lcg-lr -v --vo dteam lfn:.wn004.grid.bas.bg.0508010751
on my WN which has already taken 14:33 CPU time,
using LCG_GFAL_INFOSYS=lcg-bdii.cern.ch:2170.
The problem has been confirmed by people using different UIs from different sites (at 2_4_0).
I tried:
globus-job-run lxn1184.cern.ch /usr/bin/env qstat
Job id Name User Time Use S Queue
---------------- ---------------- ---------------- -------- - -----
4242.lxn1184 STDIN dteamsgm 00:39:30 R dteam
I think Replica Management worked a day ago, so what has happened now?
Some site polluting the BDIIs?
Something with the new glue schema?
Emanouil Atanassov
[log in to unmask]
|