Thanks for the reply Chris.
> 70 of those in today's gridftp-session.log at QMUL.
Interesting, I think this roughly corresponds to a similar rate of
errors that we see throughout all our pool nodes. However, do you know
if you have failed "important tests" because of these obnoxious
seemingly random transfer errors.
> Just a thought. Are you on the SL5 gridFTP server, or SL4 (QMUL has the infn
> sl4)?
Both, a few of our pool nodes are still running on SL4, most have
moved up to SL5. Both sets seem to be behaving the same.
> PS I can't remember whether you were copied in on the NAT issues that seem
> to affect LHCb doing gridFTP.
The SAM failures involve our poolnodes (which have internal and
external NICS) and foreign servers in the great beyond, so in theory
there shouldn't be any NAT involved. At least at our end...
cheers,
Matt
|