dCache Tier I meeting October 06, 2009

[part of a series of meetings]

Present

dCache.org(Patrick, Owen, Paul), IN2P3(), Sara(Onno), Triumf(), NDGF(Gerd), PIC(Gerard), GridKa(Doris), Fermi(Gene), CERN(Andrea)

Agenda

(see box on the other side)

Site reports

BNL

(Pedro reported via email)

• nothing to report

Fermi

Jon reported via email:

We deployed the new dccp rpm - has an env variable that allows us to  
set timeout on close.  Working so far.

Nothing else to report.

IN2P3

Lionel reported via email (copied here in case he doesn't make the meeting)

latest news from IN2P3:
- we have moved to 1.9.4-3 2 weeks ago
- we have moved to Chimera last week
So far no issues, except an issue with SRMcopy transfers: ticket #5157

(FYI, ticket ​#5157 is a problem with remote GSI-FTP transfers. The ticket is fresh, so it's too early to say what is happening here).

Sara

Onno reported that Sara is still experiencing the problem reported last week: SRM cannot handle staging of many files. Gerd has found the problem and has committed a patch to fix it. Hopefully this will fix the problem.

Onno asked when will this fix be available so they can upgrade?

We need to test before releasing, but we should have the release available for download before the next meeting.

CERN

Andrea had nothing to report from CERN.

NDGF

Downgraded over the weekend from 1.9.5 due to a deadlock, this will upgrade tomorrow.

In the meantime we seens a problem that looks like the same one Sara has been experiencing.

Other than that, everything is fine.

PIC

Things are OK, now. But, we had an issue with pin manager (reported as a RT ​ticket 4731).

Gerard is planning to upgrade to a dCache v1.9.5 on October 20th.

Patrick reported that, once we're sure the two problems (SRM staging problems; SRM deadlock problem) are fixed then we'll have the 1.9.5-2 release. NDGF will put this on their production system (tomorrow afternoon). We will watch how this release behaves under production loads, which should give you an indication whether the release is "good".

In addition, Lionel may help with the testing of 1.9.5 branch, since they have a good test framework for testing dCache releases.

Fermi

Gene had nothing more to add to Jon's report.

FZK

Doris reported that overall things are OK.

ATLAS people are pushing FZK because of the hanging dcap movers issue.

Patrick said sorry for not responding. We think we have a good idea what is happening and are discussing a fix for it.

Doris felt there was nothing else worth mentioning at the meeting.

Migration

DESY reiterated IN2P3's report: IN2P3 were able to successfully migrate from PNFS to Chimera. The process was pretty smooth and took some 72 hours to complete.

DTNM

Proposed: same time, next week.