[ https://issues.apache.org/jira/browse/ASTERIXDB-2421?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16554710#comment-16554710 ]
Chen Luo commented on ASTERIXDB-2421: ------------------------------------- This bug should have been fixed by a later patch, but Cloudberry is using a relative old version. The old issue was reported here https://issues.apache.org/jira/browse/ASTERIXDB-2309, and the fix is in https://asterix-gerrit.ics.uci.edu/#/c/2437/. This bug happens if there were multiple partitions in a single node; during recovery, previously we didn't check the partition of a FLUSH record properly. > Recovery fails with component ID mismatch > ------------------------------------------ > > Key: ASTERIXDB-2421 > URL: https://issues.apache.org/jira/browse/ASTERIXDB-2421 > Project: Apache AsterixDB > Issue Type: Bug > Reporter: Ian Maxon > Assignee: Chen Luo > Priority: Major > > It seems local recovery can fail based on the local component IDs appearing > to be ahead of the redo id somehow: > > java.lang.IllegalStateException: Illegal state of component Id. Max disk > component Id [1532110283703,1532129243741] should be less than redo flush > component Id [1532129243740,1532129243740] > at > org.apache.asterix.app.nc.RecoveryManager.redoFlush(RecoveryManager.java:797) > ~[asterix-app-0.9.4-SNAPSHOT.jar:0.9.4-SNAPSHOT] -- This message was sent by Atlassian JIRA (v7.6.3#76005)