[
https://issues.apache.org/jira/browse/HDFS-13821?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16577855#comment-16577855
]
Fei Hui commented on HDFS-13821:
--------------------------------
[~linyiqun] Test is a MR job which just does *ls* operations.
This cache is as follow
{code:java}
/** Path -> Remote location. */
private Cache<String, PathLocation> locationCache;
{code}
{code}
public PathLocation getDestinationForPath(final String path)
throws IOException {
verifyMountTable();
readLock.lock();
try {
Callable<? extends PathLocation> meh = new Callable<PathLocation>() {
@Override
public PathLocation call() throws Exception {
return lookupLocation(path);
}
};
return this.locationCache.get(path, meh);
} catch (ExecutionException e) {
throw new IOException(e);
} finally {
readLock.unlock();
}
}
{code}
Maybe it is unrelated to state store. It caches the request paths, and return
the remote location. Maybe request paths are hundreds of millions and the cache
is inefficient
> RBF:add dfs.federation.router.mount-table.cache.enable so that users can
> disable cache
> --------------------------------------------------------------------------------------
>
> Key: HDFS-13821
> URL: https://issues.apache.org/jira/browse/HDFS-13821
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: hdfs
> Affects Versions: 3.1.0, 2.9.1, 3.0.3
> Reporter: Fei Hui
> Priority: Major
> Attachments: HDFS-13821.001.patch, image-2018-08-13-11-27-49-023.png
>
>
> When i test rbf, if found performance problem.
> I found that ProxyAvgTime From Ganglia is so high, i run jstack on Router and
> get the following stack frames
> {quote}
> java.lang.Thread.State: WAITING (parking)
> at sun.misc.Unsafe.park(Native Method)
> - parking to wait for <0x00000005c264acd8> (a
> java.util.concurrent.locks.ReentrantLock$NonfairSync)
> at java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.parkAndCheckInterrupt(AbstractQueuedSynchronizer.java:836)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.acquireQueued(AbstractQueuedSynchronizer.java:870)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.acquire(AbstractQueuedSynchronizer.java:1199)
> at
> java.util.concurrent.locks.ReentrantLock$NonfairSync.lock(ReentrantLock.java:209)
> at
> java.util.concurrent.locks.ReentrantLock.lock(ReentrantLock.java:285)
> at
> com.google.common.cache.LocalCache$Segment.lockedGetOrLoad(LocalCache.java:2249)
> at
> com.google.common.cache.LocalCache$Segment.get(LocalCache.java:2228)
> at com.google.common.cache.LocalCache.get(LocalCache.java:3965)
> at
> com.google.common.cache.LocalCache$LocalManualCache.get(LocalCache.java:4764)
> at
> org.apache.hadoop.hdfs.server.federation.resolver.MountTableResolver.getDestinationForPath(MountTableResolver.java:380)
> at
> org.apache.hadoop.hdfs.server.federation.router.RouterRpcServer.getLocationsForPath(RouterRpcServer.java:2104)
> at
> org.apache.hadoop.hdfs.server.federation.router.RouterRpcServer.getLocationsForPath(RouterRpcServer.java:2087)
> at
> org.apache.hadoop.hdfs.server.federation.router.RouterRpcServer.getListing(RouterRpcServer.java:1050)
> at
> org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.getListing(ClientNamenodeProtocolServerSideTranslatorPB.java:640)
> at
> org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$2.callBlockingMethod(ClientNamenodeProtocolProtos.java)
> at
> org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:616)
> at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:982)
> at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2115)
> at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2111)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> {quote}
> Many threads blocked on *LocalCache*
> After disable the cache, ProxyAvgTime is down as follow showed
> !image-2018-08-13-11-27-49-023.png!
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
