You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hbase.apache.org by "Jim Kellerman (JIRA)" <ji...@apache.org> on 2009/03/06 21:25:56 UTC
[jira] Updated: (HBASE-1169) When a shutdown is requested, stop
scanning META regions immediately
[ https://issues.apache.org/jira/browse/HBASE-1169?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jim Kellerman updated HBASE-1169:
---------------------------------
Fix Version/s: (was: 0.19.2)
0.19.1
Assignee: Jim Kellerman
Summary: When a shutdown is requested, stop scanning META regions immediately (was: During shutdown, odd handling of regionserver hosting META)
The root cause of this problem is that the META scanners ran before all the servers were quiesced and started assigning regions that had been quiesced.
> When a shutdown is requested, stop scanning META regions immediately
> --------------------------------------------------------------------
>
> Key: HBASE-1169
> URL: https://issues.apache.org/jira/browse/HBASE-1169
> Project: Hadoop HBase
> Issue Type: Bug
> Affects Versions: 0.19.0, 0.20.0
> Reporter: Jonathan Gray
> Assignee: Jim Kellerman
> Priority: Critical
> Fix For: 0.19.1
>
>
> During shutdown of cluster, half way through quiescing servers there is a META scan in the master. The regions from servers whose leases are already canceled show up as invalid. (72.34.249.208 is hosting META)
> {code}
> 2009-01-31 10:25:42,571 INFO org.apache.hadoop.hbase.master.HMaster: Cluster shutdown requested. Starting to quiesce servers
> 2009-01-31 10:25:45,868 INFO org.apache.hadoop.hbase.master.ServerManager: Cancelling lease for 72.34.249.211:60020
> 2009-01-31 10:25:45,868 INFO org.apache.hadoop.hbase.master.ServerManager: Region server 72.34.249.211:60020: MSG_REPORT_EXITING -- lease cancelled
> 2009-01-31 10:25:47,480 INFO org.apache.hadoop.hbase.master.ServerManager: Cancelling lease for 72.34.249.216:60020
> 2009-01-31 10:25:47,480 INFO org.apache.hadoop.hbase.master.ServerManager: Region server 72.34.249.216:60020: MSG_REPORT_EXITING -- lease cancelled
> 2009-01-31 10:25:47,840 INFO org.apache.hadoop.hbase.master.ServerManager: Region server 72.34.249.210:60020 quiesced
> 2009-01-31 10:25:47,944 INFO org.apache.hadoop.hbase.master.ServerManager: Cancelling lease for 72.34.249.215:60020
> 2009-01-31 10:25:47,944 INFO org.apache.hadoop.hbase.master.ServerManager: Region server 72.34.249.215:60020: MSG_REPORT_EXITING -- lease cancelled
> 2009-01-31 10:25:48,403 INFO org.apache.hadoop.hbase.master.ServerManager: Cancelling lease for 72.34.249.213:60020
> 2009-01-31 10:25:48,403 INFO org.apache.hadoop.hbase.master.ServerManager: Region server 72.34.249.213:60020: MSG_REPORT_EXITING -- lease cancelled
> 2009-01-31 10:25:49,378 INFO org.apache.hadoop.hbase.master.ServerManager: Region server 72.34.249.218:60020 quiesced
> 2009-01-31 10:25:50,465 INFO org.apache.hadoop.hbase.master.ServerManager: Cancelling lease for 72.34.249.214:60020
> 2009-01-31 10:25:50,465 INFO org.apache.hadoop.hbase.master.ServerManager: Region server 72.34.249.214:60020: MSG_REPORT_EXITING -- lease cancelled
> 2009-01-31 10:25:59,531 INFO org.apache.hadoop.hbase.master.BaseScanner: RegionManager.metaScanner scanning meta region {regionname: .META.,,1, startKey: <>, server: 72.34.249.218:60020}
> 2009-01-31 10:25:59,544 DEBUG org.apache.hadoop.hbase.master.BaseScanner: Current assignment of activitydupehash,,1229364212541 is not valid; Server '72.34.249.214:60020' unknown.
> 2009-01-31 10:25:59,545 DEBUG org.apache.hadoop.hbase.master.BaseScanner: Current assignment of api,,1229364235220 is not valid; Server '72.34.249.216:60020' unknown.
> 2009-01-31 10:25:59,552 DEBUG org.apache.hadoop.hbase.master.BaseScanner: Current assignment of apps,,1229364222879 is not valid; Server '72.34.249.215:60020' unknown.
> 2009-01-31 10:25:59,552 DEBUG org.apache.hadoop.hbase.master.BaseScanner: Current assignment of assigners,,1229364037757 is not valid; Server '72.34.249.214:60020' unknown.
> 2009-01-31 10:25:59,554 DEBUG org.apache.hadoop.hbase.master.BaseScanner: Current assignment of canoncache,,1229364041955 is not valid; Server '72.34.249.215:60020' unknown.
> 2009-01-31 10:25:59,555 DEBUG org.apache.hadoop.hbase.master.BaseScanner: Current assignment of chunks,,1229390225893 is not valid; Server '72.34.249.211:60020' unknown.
> {code}
> Shutdown then continues as the last servers are quiesced, but at the same time the Master expires the lease on the regionserver that was hosting META and that it just scanned. It then starts to replay the logs for that regionserver in the middle of the shutdown.
> {code}
> 2009-01-31 10:25:59,799 INFO org.apache.hadoop.hbase.master.BaseScanner: RegionManager.metaScanner scan of 512 row(s) of meta region {regionname: .META.,,1, startKey: <>, server: 72.34.249.218:60020}
> complete
> 2009-01-31 10:25:59,799 INFO org.apache.hadoop.hbase.master.BaseScanner: All 1 .META. region(s) scanned
> 2009-01-31 10:26:59,530 INFO org.apache.hadoop.hbase.master.BaseScanner: RegionManager.metaScanner scanning meta region {regionname: .META.,,1, startKey: <>, server: 72.34.249.218:60020}
> 2009-01-31 10:26:59,720 INFO org.apache.hadoop.hbase.master.BaseScanner: RegionManager.metaScanner scan of 512 row(s) of meta region {regionname: .META.,,1, startKey: <>, server: 72.34.249.218:60020}
> complete
> 2009-01-31 10:26:59,720 INFO org.apache.hadoop.hbase.master.BaseScanner: All 1 .META. region(s) scanned
> 2009-01-31 10:27:40,374 INFO org.apache.hadoop.hbase.master.ServerManager: 72.34.249.218:60020 lease expired
> 2009-01-31 10:27:40,375 DEBUG org.apache.hadoop.hbase.master.HMaster: Processing todo: ProcessServerShutdown of 72.34.249.218:60020
> 2009-01-31 10:27:40,375 INFO org.apache.hadoop.hbase.master.RegionServerOperation: process shutdown of server 72.34.249.218:60020: logSplit: false, rootRescanned: false, numberOfMetaRegions: 1, onlin
> eMetaRegions.size(): 1
> 2009-01-31 10:27:40,387 INFO org.apache.hadoop.hbase.regionserver.HLog: Splitting 44 log(s) in hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020
> 2009-01-31 10:27:40,387 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Splitting 1 of 44: hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1232996040603
> 2009-01-31 10:27:40,443 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Creating new log file writer for path hdfs://mb0:9000/hbase/.META./1028785192/oldlogfile.log and region .META.,,1
> 2009-01-31 10:27:40,575 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Creating new log file writer for path hdfs://mb0:9000/hbase/sources/671225115/oldlogfile.log and region sources,,1229364117966
> 2009-01-31 10:27:41,171 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Applied 100003 total edits from hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1232996040603
> 2009-01-31 10:27:41,173 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Splitting 2 of 44: hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233093726382
> 2009-01-31 10:27:41,429 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Creating new log file writer for path hdfs://mb0:9000/hbase/dupehash/1607532582/oldlogfile.log and region dupehash,O��<L;h�,12
> 31779694744
> 2009-01-31 10:27:41,462 INFO org.apache.hadoop.hbase.master.ServerManager: 72.34.249.217:60020 lease expired
> 2009-01-31 10:27:41,499 INFO org.apache.hadoop.hbase.master.ServerManager: All user tables quiesced. Proceeding with shutdown
> 2009-01-31 10:27:41,499 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling root scanner to stop
> 2009-01-31 10:27:41,499 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling meta scanner to stop
> 2009-01-31 10:27:41,499 DEBUG org.apache.hadoop.hbase.master.RegionManager: meta and root scanners notified
> 2009-01-31 10:27:41,499 INFO org.apache.hadoop.hbase.master.RootScanner: RegionManager.rootScanner exiting
> 2009-01-31 10:27:41,499 INFO org.apache.hadoop.hbase.master.MetaScanner: RegionManager.metaScanner exiting
> 2009-01-31 10:27:41,780 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Applied 100001 total edits from hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233093726382
> 2009-01-31 10:27:41,781 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Splitting 3 of 44: hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233101015153
> 2009-01-31 10:27:41,838 INFO org.apache.hadoop.hbase.master.ServerManager: 72.34.249.210:60020 lease expired
> 2009-01-31 10:27:41,866 INFO org.apache.hadoop.hbase.master.ServerManager: All user tables quiesced. Proceeding with shutdown
> 2009-01-31 10:27:41,866 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling root scanner to stop
> 2009-01-31 10:27:41,866 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling meta scanner to stop
> 2009-01-31 10:27:41,866 DEBUG org.apache.hadoop.hbase.master.RegionManager: meta and root scanners notified
> 2009-01-31 10:27:42,557 INFO org.apache.hadoop.hbase.master.ServerManager: 72.34.249.212:60020 lease expired
> 2009-01-31 10:27:42,581 INFO org.apache.hadoop.hbase.master.ServerManager: All user tables quiesced. Proceeding with shutdown
> 2009-01-31 10:27:42,581 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling root scanner to stop
> 2009-01-31 10:27:42,581 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling meta scanner to stop
> 2009-01-31 10:27:42,581 DEBUG org.apache.hadoop.hbase.master.RegionManager: meta and root scanners notified
> 2009-01-31 10:27:42,615 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Applied 100002 total edits from hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233101015153
> 2009-01-31 10:27:42,618 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Splitting 4 of 44: hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233111791302
> 2009-01-31 10:27:43,356 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Applied 100001 total edits from hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233111791302
> 2009-01-31 10:27:43,359 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Splitting 5 of 44: hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233122447841
> 2009-01-31 10:27:43,404 INFO org.apache.hadoop.hbase.master.ServerManager: All user tables quiesced. Proceeding with shutdown
> 2009-01-31 10:27:43,404 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling root scanner to stop
> 2009-01-31 10:27:43,404 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling meta scanner to stop
> 2009-01-31 10:27:43,404 DEBUG org.apache.hadoop.hbase.master.RegionManager: meta and root scanners notified
> 2009-01-31 10:27:43,991 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Applied 100001 total edits from hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233122447841
> {code}
> During the log replay, a log file was missing from HDFS. Not sure why, there was a Datanode crash that could be related. More importantly, once it trips on the missing file it stops the replay (even though there's another 37 logs).
> {code}
> 2009-01-31 10:27:43,992 DEBUG org.apache.hadoop.hbase.regionserver.HLog: Splitting 6 of 44: hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233132556827
> 2009-01-31 10:27:44,022 WARN org.apache.hadoop.hbase.master.HMaster: Processing pending operations: ProcessServerShutdown of 72.34.249.218:60020
> java.io.FileNotFoundException: File does not exist: hdfs://mb0:9000/hbase/log_72.34.249.218_1232996040351_60020/hlog.dat.1233132556827
> at org.apache.hadoop.hdfs.DistributedFileSystem.getFileStatus(DistributedFileSystem.java:394)
> at org.apache.hadoop.fs.FileSystem.getLength(FileSystem.java:679)
> at org.apache.hadoop.io.SequenceFile$Reader.<init>(SequenceFile.java:1417)
> at org.apache.hadoop.io.SequenceFile$Reader.<init>(SequenceFile.java:1412)
> at org.apache.hadoop.hbase.regionserver.HLog.splitLog(HLog.java:742)
> at org.apache.hadoop.hbase.regionserver.HLog.splitLog(HLog.java:705)
> at org.apache.hadoop.hbase.master.ProcessServerShutdown.process(ProcessServerShutdown.java:249)
> at org.apache.hadoop.hbase.master.HMaster.processToDoQueue(HMaster.java:427)
> at org.apache.hadoop.hbase.master.HMaster.run(HMaster.java:360)
> 2009-01-31 10:27:44,022 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling root scanner to stop
> 2009-01-31 10:27:44,022 DEBUG org.apache.hadoop.hbase.master.RegionManager: telling meta scanner to stop
> 2009-01-31 10:27:44,022 DEBUG org.apache.hadoop.hbase.master.RegionManager: meta and root scanners notified
> 2009-01-31 10:27:44,023 DEBUG org.apache.hadoop.hbase.RegionHistorian: Offlined
> 2009-01-31 10:27:44,023 INFO org.apache.hadoop.hbase.master.HMaster: Stopping infoServer
> 2009-01-31 10:27:44,023 INFO org.mortbay.util.ThreadedServer: Stopping Acceptor ServerSocket[addr=0.0.0.0/0.0.0.0,port=0,localport=60010]
> 2009-01-31 10:27:44,026 INFO org.mortbay.http.SocketListener: Stopped SocketListener on 0.0.0.0:60010
> {code}
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.