wiki:tier-one-meeting-20171121
Last modified 12 months ago Last modified on 11/21/17 15:44:50

dCache Tier I meeting November 21, 2017

[part of a series of meetings]

Present

dCache.org(Paul), IN2P3(), Sara(), Triumf(), BNL(), NDGF(Ulf), PIC(), KIT(Xavier), Fermi(), CERN()

Agenda

(see box on the other side)

Site reports

KIT

dCache is working fine.

WebDAV issue

Xavier will try to enable the slow srm-upload-dir-deletion work-around; however, the problem appears to have disappeared.

The PnfsManager slow activity logging no longer shows any operations taking longer than 1 second. Xavier had decreased the threshold to 500 ms and will disable the curl configuration that extends the expect-100 deadline to see if the WebDAV issue has similarly disappeared.

ZK node problem

The unreliable ZK node has been replaced.

Do we need to restart dCache to get the new ZK connection string? Yes.

There was also a discussion on core domains and whether it makes sense to have more than one.

Ulf suggesting running core domains without any services, as this solves the start-up time. Xavier will investigate this option further.

RPM stopping dCache

Why is this?

OpenJDK reacts badly if a jar file disappears while it is running. Therefore, if dCache is upgraded, it is safer to stop dCache before upgrading.

The same is true for OpenJDK: it is safest to stop dCache before upgrading the java version.

Restart question

KIT tried to enable IPv6, but failed. The likely cause of this failure was external to dCache -- the networking infrastructure was not correctly prepared.

The symptoms were that, while the tests showed everything was fine and FTS was initially working, after a while file transfers (managed through SRM) were no longer functional.

Do we need to restart dCache to take advantage of new (IPv6) addresses? No.

lost+found

OK to just delete the entries.

Closing tickets

There has been no progress on discovering whether Xavier can close tickets that are KIT-specific, but not opened by his email address.

xrootd

The xrootd transfer errors are continuing. Jürgen is now investigating.

NDGF

Ulf had to leave early.

Support tickets for discussion

[Items are added here automagically]

DTNM

Same time, next week.