dCache Tier I meeting November 29, 2011

[part of a series of meetings]

Present

dCache.org(Tigran, Paul, Tanja, Antje), Sara(Onno), NDGF(Gerd), GridKa(Doris)

Agenda

(see box on the other side)

Site reports

SARA

Things are going well.

Some questions...

SRM space-file

We've noticed that when users delete files: they are

Only the delete turns from 0-->1 a space reservation is still there. Once in a while we get a request from ATLAS: dark data. Script that compares space file with Chimera namespace. Entries that are in SRM space-manager table but not

Do you have "srm.spacemanager.enabled=yes" on the host running the dCache domain? Yes.

The broadcast cell (part of the dCacheDomain) should forward the delete message space-manager should clean the entry.

Only when the file is deleted on the pool (i.e., cleaner runs) is the entry in srm-space-manager

Look in billing: is there a cleaner trying to delete the file? Yes, the file is deleted from the pool.

Only the SRM space file that has the dark data.

Please send email to support@… to remind Gerd to investigate this.

reg

If there is recommendation for logging settings. Standard is "WARN" on most things. Usually we don't have enough information to debug problems. Usually increase to DEBUG and ask user to try again, but if we forget to switch this off then the partition fills up and stops dCache from working.

crontab: if partition is 90% full then delete log files.

Log-rotate? Yes, and we have this quite aggressively, but sometimes

Double-information: sometimes fields are repeated on the same line.

Please send examples of such a line

Context information in square brackets, sometimes log

Perhaps a timeout to reset it back to the default level.

Would be nice

info level should be usable in production but for dCache is a bit too verbose currently.

Could configure logback to send debug to separate files, and aggressively log-rotate.

DEBUG is not recommended in production.

Can do log-level more fine-grain: this is part of the log-back configuration.

Suggest looking at the configuration chapter of the log-back configuration.

Logging severity of the message? You can do this. Look in the logback.xml

Helpful to have the severity of message.

GridKa

Everything is fine so far.

Check for metadata?

NDGF

Nothing to report.

Support tickets for discussion

[Items are added here automagically]

DTNM

Same time, next week.