This page last changed on Nov 24, 2010 by kskaburs.

LFC probes

LFC probes are provided by ch.cern and org.sam.

ch.cern

ch.cern/LFC-probe is part of grid-monitoring-probes-ch.cern RPM and contains metrics to check LFC service. The probe is written in Python and uses python-GridMon library.

Metrics

Name Description
ch.cern.LFC-Read Test if an entry in the catalog can be read.
ch.cern.LFC-Write Test if the modification time of an entry in the catalog can be updated.
ch.cern.LFC-Readdir Time how long it takes to read a directory (/grid).
ch.cern.LFC-ReadDli Do a read from a DLI.
ch.cern.LFC-Ping Ping LFC service (service level ping).

org.sam

org.sam/LFC-probe is part of grid-monitoring-probes-org.sam RPM and contains a metric to do cleanup of LFC's test area mainly used by WN metrics.

Metrics

Name Description
org.sam.LFC-Cleanup Clean test area on LFC

org.sam.LFC-Cleanup

Given a directory in LFC the metric tries to delete all replicas of the files found in the current directory and then un-register the files. It's multi-threaded, works asynchronously and builds local cache (/var/lib/gridprobes/<VO or FQAN>/<namespace>/<hostname>/lfns.db - Python pickle format) of the files to be cleaned up to minimize load on LFC and speed up the cleanup process on restarts (after a definable timeout kicks in if the metric wasn't able to process all the files). Metric parameters:

--cleanup-timeout   <sec>   Cleanup timeout. (Default: 300 sec)
--cleanup-dir       <dir>   Directory to clean. (Default: /grid/<VO>/SAM)
--cleanup-file-ttl  <hrs>   Time for a file to stay on LFC. (Default: 15 days)
--cleanup-files-max <num>   Number of files to delete at most before timeout
                            kicks in. (Default: maxint)
--cleanup-threads   <num>   Number of cleanup threads. (Default: 5)
Document generated by Confluence on Feb 27, 2014 10:19