Hi everyone,
I don't know if anyone will have experienced this before, but we are
having trouble consistently passing the SFTs. We are failing on:
Checking 3rd party replication from Central SE (lxn1183.cern.ch)
https://lcg-sft.cern.ch:9443/sft/sitehistory.cgi?site=ce.epcc.ed.ac.uk
As an example:
[gcowan@ui dcache]$ lcg-cr --vo dteam -d lxn1183.cern.ch -l
lfn:cowan261005_14 file:///usr/lib/X11/rgb.txt
guid:bc9d1998-6d91-438b-a183-b7a3d3456456
[gcowan@ui dcache]$ lcg-rep -v --vo dteam -d srm.epcc.ed.ac.uk
lfn:cowan261005_14
Using grid catalog type: edg
Source URL: lfn:cowan261005_14
File size: 17371
VO name: dteam
Destination specified: srm.epcc.ed.ac.uk
Source URL for copy:
gsiftp://lxn1183.cern.ch/storage/dteam/generated/2005-10-26/file6542f4fb-583c-4113-9caa-8367dbb4d0d6
Destination URL for copy:
gsiftp://srm.epcc.ed.ac.uk:2811//pnfs/epcc.ed.ac.uk/data/dteam/generated/2005-10-26/file05d990fe-8d9e-4bc1-8092-034efaf65ec8
# streams: 1
# set timeout to 0
0 bytes 0.00 KB/sec avg 0.00 KB/sec instlcg_rep:
Transport endpoint is not connected
Inspection of the error messages from the rm sections of the above URL
shows that the failures are occuring whenever lcg-rep is passed a TURL for
the gridftp door on our admin node (srm). However, we pass the test if
our dCache returns a TURL to lcg-rep for the gridftp door on the pool
node (dcache). Inspection of the gridftp logs on the admin node
shows the following non-ending series of error messages:
10/27 16:30:44 Cell(GFTP-Unknown-113@gridftpdoorDomain) : CellAdapter:
cought SocketTimeoutException: Do nothing, just allow looping to continue
10/27 16:30:44 Cell(GFTP-Unknown-113@gridftpdoorDomain) : CellAdapter:
cought SocketTimeoutException: Do nothing, just allow looping to continue
10/27 16:30:44 Cell(GFTP-Unknown-113@gridftpdoorDomain) : CellAdapter:
cought SocketTimeoutException: Do nothing, just allow looping to continue
The same message are seen on the pool node, until a connection is accepted
from the client, at which point the transfer completes.
The behaviour seems strange since non-3rd party srmcp's and
globus-url-copies into and out of our dCache via the admin node door work
fine. Could this be due to some problem with gridftp acting in
active/passive mode? Has anyone seen anything like this before?
Cheers,
Greig
--
=======================================================================
Dr Greig A Cowan http://www.ph.ed.ac.uk/~gcowan1
School of Physics, University of Edinburgh, James Clerk Maxwell Building
TIER-2 STORAGE SUPPORT PAGES: http://wiki.gridpp.ac.uk/wiki/Grid_Storage
=======================================================================
|