Hi all,
> So, Rafal's proxy was OK to submit the job, but by the time the job tries
> to globus-url-copy the ISB from the WMS, the proxy suddenly fails.
>
> I have submitted jobs as "ops" and "ops/Role=lcgadmin" to the CE,
> using a CERN WMS and the WMS at CNAF: they all succeeded.
>
> In theory it is possible that CREAM somehow gave the job on the WN
> a bad proxy, but it looks very unlikely.
>
> The "Login incorrect" message may be misleading. The proxy on the WN
> could have a problem that causes the server to close the connection
> already at an earlier stage than the mapping.
For one example proxy we found exactly that: its VOMS attributes had
expired in the meantime, so the GridFTP server correctly rejected it.
That is a "user error" combined with a WMS bug:
https://savannah.cern.ch/bugs/index.php?28167
The patch for that has been certified, but is waiting to be released.
While the relevant WMS nodes have not all been updated with the patch,
the user must ensure the VOMS attributes have the _same_ lifetime as
the proxy itself.
|