TXLockIdImpl memory leak after CommitConflictException

Products

• VMware Tanzu Gemfire

Issue/Introduction

From the logs, we can see that commit processing was slow:

[warn 2023/06/23 11:04:24.417 EST Server1 <ServerConnection on port 43031 Thread 2182> tid=0x886] 15 seconds have elapsed while waiting for replies: <TXCommitMessage$CommitReplyProcessor 17435100 waiting for 1 replies from [xx.xx.xx.xx(Server2:xxxxx)<v46>:xxxxx]> on xx.xx.xx.xx(Server1:xxxx)<v43>:xxxxx whose current membership list is: [[xx.xx.xx.xx(Server1:xxxx)<v43>:xxxx, xx.xx.xx.xx(Server2:xxxx)<v46>:xxxx, xx.xx.xx.xx(locator-1:xxxxx:locator)<ec><v10>:xxxx, xx.xx.xx.xx(Locator1:xxxxxx:locator)<ec><v11>:xxxx, xx.xx.xx.xx(Locator2:xxxxx:locator)<ec><v0>:xxxx, xx.xx.xx.xx(Server3:xxxx)<v39>:xxxxx]

[info 2023/06/23 11:04:38.900 EST Server1 <ServerConnection on port 43031 Thread 2182> tid=0x886]  TXCommitMessage$CommitReplyProcessor wait for replies completed

Environment

Product Version: 9.15

Resolution

This could be caused by the TXLockIdImpl memory leak issue mentioned in GEODE-10395  (internal bug number GEM-4219), which is fixed in GemFire 10.0 and backported to GemFire 9.15.7.