We're updating the issue view to help you get more done. 

Loop of restarts of PU instance in same container cause memory leak

Description

steps to reproduce:
1.restart of backup due to memory shortage on same container


set gsc with low heap size (128M)
start default agent
deploy 1,1 space (max instances per vm 1)
start benchmark
wait till getting memory shortage
see that gsm keep restarting the backup and fails to deploy it due to the shortage.
wait a while and take heap dump of the container that backup was initially on, and now contains no space instance
in heap you can see lots of space proxies and several cache managers space engines all are rooted to net.jini.lease.LeaseRenewalManager
that keeps net.jini.lookup.ServiceDiscoveryManager$LeaseListenerImpl

2. loop of relocation of pu instances


The gsc heap dump for is attached to case 11037. Please download relocationStuck_cacheCloseMaxWait5.7z.001
Run Test in jira : xap-13204 with com.gs.cache_context_close_max_wait 2 sec. and take heap dump after a while.

Workaround

None

Acceptance Test

No acceptance test required

Status

Assignee

Alon Shoham

Reporter

Ester Atzmon

Labels

None

Priority

Medium

SalesForce Case ID

10618,11037

Fix versions

Commitment Version/s

None

Due date

None

Product

None

Edition

None

Platform

All