[
https://issues.apache.org/jira/browse/HBASE-21069?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16584452#comment-16584452
]
Andrew Purtell edited comment on HBASE-21069 at 8/17/18 10:24 PM:
------------------------------------------------------------------
Actually that might be the right approach.
StoreScanner#updateReaders is called from here
{code}
private void notifyChangedReadersObservers(List<StoreFile> sfs) throws
IOException {
for (ChangedReadersObserver o : this.changedReaderObservers) {
List<KeyValueScanner> memStoreScanners;
this.lock.readLock().lock();
try {
memStoreScanners = this.memstore.getScanners(o.getReadPoint());
} finally {
this.lock.readLock().unlock();
}
---> o.updateReaders(sfs, memStoreScanners);
}
}
{code}
And DefaultMemStore#getScanners can return null.
{code}
public List<KeyValueScanner> getScanners(long readPt) {
MemStoreScanner scanner =
new MemStoreScanner(activeSection, snapshotSection, readPt, comparator);
scanner.seek(CellUtil.createCell(HConstants.EMPTY_START_ROW));
if (scanner.peek() == null) {
scanner.close();
---> return null;
}
return Collections.<KeyValueScanner> singletonList(scanner);
}
{code}
was (Author: apurtell):
Actually that might not be the wrong approach.
StoreScanner#updateReaders is called from here
{code}
private void notifyChangedReadersObservers(List<StoreFile> sfs) throws
IOException {
for (ChangedReadersObserver o : this.changedReaderObservers) {
List<KeyValueScanner> memStoreScanners;
this.lock.readLock().lock();
try {
memStoreScanners = this.memstore.getScanners(o.getReadPoint());
} finally {
this.lock.readLock().unlock();
}
---> o.updateReaders(sfs, memStoreScanners);
}
}
{code}
And DefaultMemStore#getScanners can return null.
{code}
public List<KeyValueScanner> getScanners(long readPt) {
MemStoreScanner scanner =
new MemStoreScanner(activeSection, snapshotSection, readPt, comparator);
scanner.seek(CellUtil.createCell(HConstants.EMPTY_START_ROW));
if (scanner.peek() == null) {
scanner.close();
---> return null;
}
return Collections.<KeyValueScanner> singletonList(scanner);
}
{code}
> NPE in StoreScanner.updateReaders causes RS to crash
> -----------------------------------------------------
>
> Key: HBASE-21069
> URL: https://issues.apache.org/jira/browse/HBASE-21069
> Project: HBase
> Issue Type: Bug
> Affects Versions: 1.3.2
> Reporter: Thomas D'Silva
> Priority: Major
> Attachments: HBASE-21069-branch-1.patch
>
>
> I see the following NPE in the region server log for a table that is taking
> heavy writes.
> I am not sure how the {{memStoreScanners}} variable gets set to null.
> {code}
> 2018-08-17 19:59:23,682 DEBUG [MemStoreFlusher.1]
> regionserver.HRegionFileSystem - Committing store file ...
> 2018-08-17 19:59:23,684 INFO [MemStoreFlusher.1] regionserver.HStore - Added
> hdfs://...., entries=919170, sequenceid=275114, filesize=22.6 M
> 2018-08-17 19:59:23,689 FATAL [MemStoreFlusher.1] regionserver.HRegionServer
> - ABORTING region server
> iotperf1dchbase1a-dnds22-2-prd.eng.sfdc.net,60020,1533915690501: Replay of
> WAL required. Forcing server shutdown
> org.apache.hadoop.hbase.DroppedSnapshotException: region: ......
> at
> org.apache.hadoop.hbase.regionserver.HRegion.internalFlushCacheAndCommit(HRegion.java:2581)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.internalFlushcache(HRegion.java:2258)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.internalFlushcache(HRegion.java:2220)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.flushcache(HRegion.java:2106)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.flush(HRegion.java:2031)
> at
> org.apache.hadoop.hbase.regionserver.MemStoreFlusher.flushRegion(MemStoreFlusher.java:508)
> at
> org.apache.hadoop.hbase.regionserver.MemStoreFlusher.flushRegion(MemStoreFlusher.java:478)
> at
> org.apache.hadoop.hbase.regionserver.MemStoreFlusher.access$900(MemStoreFlusher.java:76)
> at
> org.apache.hadoop.hbase.regionserver.MemStoreFlusher$FlushHandler.run(MemStoreFlusher.java:264)
> at java.lang.Thread.run(Thread.java:745)
> Caused by: java.lang.NullPointerException
> at java.util.ArrayList.<init>(ArrayList.java:177)
> at
> org.apache.hadoop.hbase.regionserver.StoreScanner.updateReaders(StoreScanner.java:827)
> at
> org.apache.hadoop.hbase.regionserver.HStore.notifyChangedReadersObservers(HStore.java:1160)
> at
> org.apache.hadoop.hbase.regionserver.HStore.updateStorefiles(HStore.java:1133)
> at
> org.apache.hadoop.hbase.regionserver.HStore.access$900(HStore.java:120)
> at
> org.apache.hadoop.hbase.regionserver.HStore$StoreFlusherImpl.commit(HStore.java:2487)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.internalFlushCacheAndCommit(HRegion.java:2536)
> ... 9 more
> 2018-08-17 19:59:23,692 FATAL [MemStoreFlusher.1] regionserver.HRegionServer
> - RegionServer abort: loaded coprocessors are:
> [org.apache.hadoop.hbase.security.access.AccessController,
> org.apache.phoenix.coprocessor.ScanRegionObserver,
> org.apache.phoenix.coprocessor.UngroupedAggregateRegionObserver,
> org.apache.phoenix.hbase.index.Indexer,
> org.apache.phoenix.coprocessor.GroupedAggregateRegionObserver,
> org.apache.hadoop.hbase.security.token.TokenProvider,
> org.apache.phoenix.coprocessor.ServerCachingEndpointImpl]
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
