Apache
/
hbase
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

HBASE-9099: logReplay could trigger double region assignment 

git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1510615 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
jeffreyz 2013-08-05 17:28:13 +00:00
parent 2036dbb763
commit 96e750fdac
3 changed files with 19 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
hbase-server/src/main/java/org/apache/hadoop/hbase/master/AssignmentManager.java
View File

@ -2776,11 +2776,10 @@ public class AssignmentManager extends ZooKeeperListener {
* @param hri
* @param timeOut Milliseconds to wait for current region to be out of transition state.
* @return True when a region clears regions-in-transition before timeout otherwise false
* @throws IOException
* @throws InterruptedException
*/
public boolean waitOnRegionToClearRegionsInTransition(final HRegionInfo hri, long timeOut)
throws IOException, InterruptedException {
throws InterruptedException {
if (!regionStates.isRegionInTransition(hri)) return true;
RegionState rs = null;
long end = (timeOut <= 0) ? Long.MAX_VALUE : EnvironmentEdgeManager.currentTimeMillis()

4
hbase-server/src/main/java/org/apache/hadoop/hbase/master/handler/MetaServerShutdownHandler.java
View File

@ -85,7 +85,9 @@ public class MetaServerShutdownHandler extends ServerShutdownHandler {
if (this.shouldSplitHlog && this.distributedLogReplay) {
if (!am.waitOnRegionToClearRegionsInTransition(HRegionInfo.FIRST_META_REGIONINFO,
regionAssignmentWaitTimeout)) {
throw new IOException("Region " + HRegionInfo.FIRST_META_REGIONINFO.getEncodedName()
// Wait here is to avoid log replay hits current dead server and incur a RPC timeout
// when replay happens before region assignment completes.
LOG.warn("Region " + HRegionInfo.FIRST_META_REGIONINFO.getEncodedName()
+ " didn't complete assignment in time");
}
this.services.getMasterFileSystem().splitMetaLog(serverName);

30
hbase-server/src/main/java/org/apache/hadoop/hbase/master/handler/ServerShutdownHandler.java
View File

@ -19,6 +19,7 @@
package org.apache.hadoop.hbase.master.handler;
import java.io.IOException;
import java.io.InterruptedIOException;
import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;
@ -278,26 +279,25 @@ public class ServerShutdownHandler extends EventHandler {
throw new IOException(ie);
}
try {
if (this.shouldSplitHlog && this.distributedLogReplay) {
// wait for region assignment completes
for (HRegionInfo hri : toAssignRegions) {
if (this.shouldSplitHlog && this.distributedLogReplay) {
// wait for region assignment completes
for (HRegionInfo hri : toAssignRegions) {
try {
if (!am.waitOnRegionToClearRegionsInTransition(hri, regionAssignmentWaitTimeout)) {
throw new IOException("Region " + hri.getEncodedName()
// Wait here is to avoid log replay hits current dead server and incur a RPC timeout
// when replay happens before region assignment completes.
LOG.warn("Region " + hri.getEncodedName()
+ " didn't complete assignment in time");
}
} catch (InterruptedException ie) {
throw new InterruptedIOException("Caught " + ie
+ " during waitOnRegionToClearRegionsInTransition");
}
// submit logReplay work
this.services.getExecutorService().submit(
new LogReplayHandler(this.server, this.services, this.deadServers, this.serverName));
hasLogReplayWork = true;
}
} catch (Exception ex) {
if (ex instanceof IOException) {
resubmit(serverName, (IOException)ex);
} else {
throw new IOException(ex);
}
// submit logReplay work
this.services.getExecutorService().submit(
new LogReplayHandler(this.server, this.services, this.deadServers, this.serverName));
hasLogReplayWork = true;
}
} finally {
this.deadServers.finish(serverName);