The JDK bug report I submitted on this issue has been accepted and it is under review:
https://bugs.openjdk.java.net/browse/JDK-8079274 Here is what the ParNewGC pauses look like in the Kafka logs in one of our environments (earlier I provided logs from the test program). In this case I had already increased the ZooKeeper session timeout to 20 seconds so that it did not trigger a communication failure while waiting for GC to finish. [2015-04-24 13:26:23,244] INFO 0 successfully elected as leader (kafka.server.ZookeeperLeaderElector) 2.111: [GC (Allocation Failure) 2.111: [ParNew: 136320K->10236K(153344K), 0.0235777 secs] 648320K->522236K(2080128K), 0.0237092 secs] [Times: user=0.03 sys=0.01, real=0.02 secs] 2.599: [GC (Allocation Failure) 2.599: [ParNew: 146556K->3201K(153344K), 9.1514626 secs] 658556K->519191K(2080128K), 9.1515757 secs] [Times: user=18.25 sys=0.01, real=9.15 secs] [2015-04-24 13:26:33,443] INFO New leader is 0 (kafka.server.ZookeeperLeaderElector$LeaderChangeListener) James Lent Senior Software Engineer Digitalsmiths A TiVo Company www.digitalsmiths.com<http://www.digitalsmiths.com/> jl...@digitalsmiths.com<mailto:jl...@digitalsmiths.com> | office 919.460.4747 ________________________________ This email and any attachments may contain confidential and privileged material for the sole use of the intended recipient. Any review, copying, or distribution of this email (or any attachments) by others is prohibited. If you are not the intended recipient, please contact the sender immediately and permanently delete this email and any attachments. No employee or agent of TiVo Inc. is authorized to conclude any binding agreement on behalf of TiVo Inc. by email. Binding agreements with TiVo Inc. may only be made by a signed written agreement.