-
Bug
-
Resolution: Done
-
Major
-
JBoss AMQ 6.3 R15
-
None
NPE is observed with mKahaDB and enableIndexWriteAsync="true". Currently determining if using the default "false" value for enableIndexWriteAsync mitigates the issue.
2020-08-30 07:11:53,853 | INFO | haDB Page Writer | PageFile | .store.kahadb.disk.page.PageFile 1113 | 162 - org.apache.activemq.activemq-osgi - 5.11.0.redhat-630434 | An exception was raised while performing poll writes
java.io.IOException: Cannot create recovery file. Reason: java.lang.NullPointerException
at org.apache.activemq.util.IOExceptionSupport.create(IOExceptionSupport.java:28)[162:org.apache.activemq.activemq-osgi:5.11.0.redhat-630434]
at org.apache.activemq.store.kahadb.disk.page.PageFile.writeBatch(PageFile.java:1154)[162:org.apache.activemq.activemq-osgi:5.11.0.redhat-630434]
at org.apache.activemq.store.kahadb.disk.page.PageFile.pollWrites(PageFile.java:1110)[162:org.apache.activemq.activemq-osgi:5.11.0.redhat-630434]
at org.apache.activemq.store.kahadb.disk.page.PageFile.access$300(PageFile.java:67)[162:org.apache.activemq.activemq-osgi:5.11.0.redhat-630434]
at org.apache.activemq.store.kahadb.disk.page.PageFile$3.run(PageFile.java:1310)[162:org.apache.activemq.activemq-osgi:5.11.0.redhat-630434]
Caused by: java.lang.NullPointerException
at java.util.zip.Adler32.update(Adler32.java:73)[:1.8.0_252]
at org.apache.activemq.store.kahadb.disk.page.PageFile.writeBatch(PageFile.java:1152)[162:org.apache.activemq.activemq-osgi:5.11.0.redhat-630434]
... 3 more
- When this happens, the only known way to recover is to remove the entire kahadb (not just delete the index files).
- If you don't remove the KahaDB, the NPE is hit immediately after broker restart and the broker becomes unresponsive.
- When KahaDB is removed, the issue is resolved
