In the event of an OOME exception, the dumps should be written to a suitable location so we can investigate. -XX:HeapDumpPath=/tmp would be good, but until we repartition cm-hadoop06, that could be trouble for it since it only has 500MB free on its root partition.
Daniel, Do you think /data1/hadoop would be okay for the prod machines? They all have that directory and it would have the space.
Yep, that is fine.
This has been checked in and deployed. The services have *NOT* been restarted yet so this change won't take effect until then. I didn't want to create unnecessary downtime for this though and figured this can take effect the next time we have a restart.
Status: NEW → RESOLVED
Last Resolved: 8 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.