What's the best way to parse GC logs?

[ddrillic]

We would like Splunk to automatically parse our GC logs and we found a wiki link on configuring props.conf to do so at Community:Monitoring JVMs
HOWTO: Index your JVM garbage collection data

Should we do it at index time or search time?

[somesoni2 · Accepted solution]

From that link

On Indexers (or heavy forwarders)
props.conf

[sun_jvm]
AUTO_LINEMERGE=FALSE
SHOULD_LINEMERGE=TRUE
DATETIME_CONFIG=CURRENT
BREAK_ONLY_BEFORE=\d+\.\d+:

On search heads
props.conf

[sun_jvm]
REPORT-jvm = sun_jvm_gc

transforms.conf

[sun_jvm_gc]
REGEX = \[(Full\s)?GC\s(?<JVM_HeapUsedBeforeGC>\d+)K->(?<JVM_HeapUsedAfterGC>\d+)K\((?<JVM_HeapSize>\d+)K\),\s(?<JVM_GCTimeTaken>\d+.\d+)\ssecs\]

[ddrillic]

Gorgeous @somesoni2 - you know, our sales engineer says - Always at search time - what do you think?

[somesoni2]

He must be talking about field extractions which are recommended during search-time. Above config does exactly that.

There are cases where we do index-time field extractions, e.g. for data which we want to use data model acceleration or want to use the fields in tstats. They occupy space on indexers (and will be replicated in case of indexer clusters) and will cause slightly slowness in indexing process, hence they should be avoided unless you're OK with that impact.

[ddrillic]

Gorgeous @somesoni2 !!!