[ https://issues.apache.org/jira/browse/HBASE-14450?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15427299#comment-15427299 ]
Ted Yu commented on HBASE-14450: -------------------------------- With more debug logging, the incremental restore failure can be attributed to missed log archiving due to potential accounting error. See the following snippet w.r.t. 10.22.9.171%2C63282%2C1471559038490.regiongroup-3.1471559070013, the wal file written with 5 incremental rows: {code} 2016-08-18 15:24:57,928 DEBUG [sync.3] wal.FSHLog$SyncRunner(1277): syncing writer hdfs://localhost:63272/user/tyu/test-data/2479f191-cda8-4ea6-a865-15612aee031a/WALs/10.22.9.171,63282,1471559038490/10.22.9.171%2C63282%2C1471559038490.regiongroup-3.1471559070013 2016-08-18 15:24:57,943 DEBUG [main] backup.TestIncrementalBackup(87): written 5 rows to ns2:test-14715590609531 ... 2016-08-18 15:24:58,981 DEBUG [regionserver//10.22.9.171:0.logRoller] wal.SequenceIdAccounting(335): not archiving hdfs://localhost:63272/user/tyu/test-data/2479f191-cda8-4ea6-a865- 15612aee031a/WALs/10.22.9.171,63282,1471559038490/10.22.9.171%2C63282%2C1471559038490.regiongroup-3.1471559070013 value: 110 oldestFlushing: 9223372036854775807 oldestUnflushed: 106 {code} > HBase Backup/Restore Phase 3: Multiwal support > ---------------------------------------------- > > Key: HBASE-14450 > URL: https://issues.apache.org/jira/browse/HBASE-14450 > Project: HBase > Issue Type: Task > Affects Versions: 2.0.0 > Reporter: Vladimir Rodionov > Assignee: Ted Yu > Labels: backup > Fix For: 2.0.0 > > Attachments: 14450-roll.v2.txt, 14450.roll-wait.txt, 14450.v2.txt, > org.apache.hadoop.hbase.backup.TestIncrementalBackup-output.txt, > testIncrementalBackup-8-15.txt, testIncrementalBackup-succ.txt > > > We need to support multiwal configurations. -- This message was sent by Atlassian JIRA (v6.3.4#6332)