IC's RB's II looks at all the manually configured individual griises but
for Lawrence's sake this is still the reason why matching on
the IC RB is not as good as matching on the cern BD-II.
Steve
On Mon, 17 Mar 2003, Stephen Burke wrote:
> On Mon, 17 Mar 2003, Lawrence Lowe wrote:
> > Of the 49 jobs via the IC RB, 34 matched successfully to BHAM, but 15 got
> > status Aborted/No matching resource found. These were randomly
> > interspersed with the good submits. Most of the 34 arrived here within
> > 10-12 minutes, apart from 5 whose delay ranged from 20 minutes to
> > 140 minutes(!).
>
> Is the IC RB still using an MDS-based II rather than a DBII? The "no
> matching resource" response is a symptom of that, the underlying top-level
> MDS doesn't respond within the II timeout so it sees no information.
> (By contrast the DBII always sees information but it may be very stale.)
>
> Also, if that's the case, is the II pointed at the new RAL MDS? That
> might explain why it seems so slow ...
>
> In any case any RB should be using a BDII, MDS really doesn't cope!
>
> Stephen
>
--
Steve Traylen
[log in to unmask]
http://www.gridpp.ac.uk/
|