Testbed Support for GridPP member institutes
> [mailto:[log in to unmask]] On Behalf Of Winnie Lacesso said:
> Bristol is seeing a lot lately, both on HPC & PP clusters, of hanging
> lcg-cp where the user has set a timeout, but the timeout is
> not working.
One thing which might be of interest is that there's a change in the
works to have more flexible timeouts in lcg-utils, although it may not
help with this:
https://savannah.cern.ch/bugs/?42517
The related patch seems to still be in certification:
https://savannah.cern.ch/patch/?2783
> Why isn't the lcg-cp timeout working for the user's grid job?
> Or is this a bug?
It could be a bug, or just that it's hanging in something which can't be
interrupted at some low level. If it's made a zero-length file I would
somewhat suspect a failure on the gridftp data channel, e.g. a mismatch
between the port ranges that means the data channel sometimes doesn't
connect depending on which port it picks - but I could be completely
wrong. You could ask the remote site which ports they have open in their
firewall ...
Stephen
--
Scanned by iCritical.
|