Some comments on the minutes:
1. Quick site upgrade status
Deploying the DPM security patch - getting urgent: According to the
mysteriously vanishing email from Kostas (Jens hadn't received it
but everybody else had), the GridFTP server fix does not fix the
problem. It is now necessary to put it back into the loop,
involving the EGEE vulnerabilities group properly, this time.
ACTION Graeme - follow up, Jens follow up with Linda Cornwall.
We now have some more information on this. Will feed this back soon.
DPM recommendations: Jens asked whether sites on 1.5.10 should
upgrade now to 1.6.3 and from there to 1.6.5, or they should wait
for someone to implement the upgrade path directly from 1.5.10 to
1.6.5? Graeme suggests that even the gLite 20 YAIM which upgrades
from 1.5.10 to 1.6.3 involves schema changes, and people should
perhaps extract the relevant code from the later gLite release and
run it separately (because the full YAIM codeset may do other
things to the DPM that you don't want done).
So in conclusion, the bug is still open, and sites should get in
touch with Graeme and Greig about the DPM upgrade, which is likely
to require some testing/documenting/feedback/handholding/support.
1.5.10 -> 1.6.3 -> 1.6.5 (all using YAIM) will work. This is what we did
at Cambridge (with an intermediate 1.6.4) and everything seems to be OK.
And 1.8.0 testing... Greig may have one but Lancaster is on
1.7.0-38 and other dCache sites are on 1.7.0-X as well. The 1.8.0
release is not considered ready for production yet, but Owen is
very interested in feedback on the release.
See my email from yesterday with comments on this stuff. Greig
definitely has one. Other sites shouldn't, unless they've got a spare
test machine that's sitting about doing nothing.
2. Quick storage publishing discussion
So Jens had remarked that currently everybody in UKI is publishing
zero, *except* JET, so the details of the published data could not
be discussed. Nobody knew the cause of this, so it's probably just
a bug. ACTION Jens report it.
Are you talking about the storage accounting page? Dave Kant infomred me
today that there was a problem with the filesystem of the server,
resulting in some things breaking.
Greig
Jensen, J (Jens) wrote:
> Minutes have now been uploaded to the agenda page:
>
> http://indico.cern.ch/conferenceDisplay.py?confId=18463
>
> There was less discussion of the glue issues, so we finished
> a bit early. Which is OK of course.
>
> I am now seeing the mails again from Kostas... weird why
> I didn't see them the first time around, maybe a synchronisation
> problem.
>
> Cheers
> --jens
|