On Wed, 15 Nov 2006, Gonzalo Merino wrote:
> Hello lcg-CE guru,
>
> Ware having some jobs running in some WNs that seems to be "stuck". They
> are not consuming cpu at all for some reason we do not quite understand.
>
> I have attached at the end of the mail the processes I can see running
> at this moment on the WN for the given user. There, I see that the
> process that it seems is running is a perl one doing some "maradona"
> things... may be the wrapper trying to contact back the RB or something
> like this.
>
> Is there a way I can know what is this job really doing? How to debug
> wether it is a problem of the wrapper, in trying to send back some
> output sandbox or something like this?
The wrapper will try to upload the output sandbox and the Maradona file:
if a globus-url-copy fails for any reason, the operation is retried after
a delay that gets doubled every time, up to a maximum of 5 hours.
This is to allow for a temporary network problem or downtime of the RB.
|