[CASSANDRA-10815] WrappingCompactionStrategy can block others when there are lots of sstables

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Duplicate
• Component/s: Local/Compaction
• Severity: Normal
• Assignee: Marcus Eriksson
• Reporter: Severin Leonhardt

Authors

Marcus Eriksson

• Created: 04/Dec/15 10:45
• Updated: 16/Apr/19 09:30
• Resolved: 04/Dec/15 13:07

Description

We observe nodetool compactionstats hanging when there are a lot of SSTables in one table. We have about 30.000 SSTables most likely created by an incremental repair (why it's that many is still a mystery to us).

Looking at the stacktraces of some selected threads it becomes apparent that a single CompactionExecutor blocks several other threads:

"CompactionExecutor:4065" - Thread t@282454
   java.lang.Thread.State: RUNNABLE
	at java.util.TimSort.binarySort(TimSort.java:292)
	at java.util.TimSort.sort(TimSort.java:235)
	at java.util.Arrays.sort(Arrays.java:1512)
	at java.util.ArrayList.sort(ArrayList.java:1454)
	at java.util.Collections.sort(Collections.java:175)
	at org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy.trimToThresholdWithHotness(SizeTieredCompactionStrategy.java:139)
	at org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy.mostInterestingBucket(SizeTieredCompactionStrategy.java:119)
	at org.apache.cassandra.db.compaction.LeveledManifest.getSSTablesForSTCS(LeveledManifest.java:360)
	at org.apache.cassandra.db.compaction.LeveledManifest.getCompactionCandidates(LeveledManifest.java:318)
	- locked <25446e7b> (a org.apache.cassandra.db.compaction.LeveledManifest)
	at org.apache.cassandra.db.compaction.LeveledCompactionStrategy.getMaximalTask(LeveledCompactionStrategy.java:101)
	at org.apache.cassandra.db.compaction.LeveledCompactionStrategy.getNextBackgroundTask(LeveledCompactionStrategy.java:90)
	- locked <33506169> (a org.apache.cassandra.db.compaction.LeveledCompactionStrategy)
	at org.apache.cassandra.db.compaction.WrappingCompactionStrategy.getNextBackgroundTask(WrappingCompactionStrategy.java:77)
	- locked <31924612> (a org.apache.cassandra.db.compaction.WrappingCompactionStrategy)
	at org.apache.cassandra.db.compaction.CompactionManager$BackgroundCompactionCandidate.run(CompactionManager.java:230)
	at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

Because this thread has locked <31924612> (a org.apache.cassandra.db.compaction.WrappingCompactionStrategy) the following threads are blocked:

"CompactionExecutor:4064" - Thread t@282337
   java.lang.Thread.State: BLOCKED
	at org.apache.cassandra.db.compaction.WrappingCompactionStrategy.getNextBackgroundTask(WrappingCompactionStrategy.java:72)
	- waiting to lock <31924612> (a org.apache.cassandra.db.compaction.WrappingCompactionStrategy) owned by "CompactionExecutor:4065" t@282454
	at org.apache.cassandra.db.compaction.CompactionManager$BackgroundCompactionCandidate.run(CompactionManager.java:230)
	at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

"MemtableFlushWriter:2431" - Thread t@283411
   java.lang.Thread.State: BLOCKED
	at org.apache.cassandra.db.compaction.WrappingCompactionStrategy.handleNotification(WrappingCompactionStrategy.java:250)
	- waiting to lock <31924612> (a org.apache.cassandra.db.compaction.WrappingCompactionStrategy) owned by "CompactionExecutor:4065" t@282454
	at org.apache.cassandra.db.DataTracker.notifyAdded(DataTracker.java:518)
	at org.apache.cassandra.db.DataTracker.replaceFlushed(DataTracker.java:178)
	at org.apache.cassandra.db.compaction.AbstractCompactionStrategy.replaceFlushed(AbstractCompactionStrategy.java:234)
	at org.apache.cassandra.db.ColumnFamilyStore.replaceFlushed(ColumnFamilyStore.java:1541)
	at org.apache.cassandra.db.Memtable$FlushRunnable.runMayThrow(Memtable.java:336)
	at org.apache.cassandra.utils.WrappedRunnable.run(WrappedRunnable.java:28)
	at com.google.common.util.concurrent.MoreExecutors$SameThreadExecutorService.execute(MoreExecutors.java:297)
	at org.apache.cassandra.db.ColumnFamilyStore$Flush.run(ColumnFamilyStore.java:1154)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

"RMI TCP Connection(128577)-192.168.0.1" - Thread t@282977
   java.lang.Thread.State: BLOCKED
	at org.apache.cassandra.db.compaction.WrappingCompactionStrategy.getEstimatedRemainingTasks(WrappingCompactionStrategy.java:147)
	- waiting to lock <31924612> (a org.apache.cassandra.db.compaction.WrappingCompactionStrategy) owned by "CompactionExecutor:4065" t@282454
	at org.apache.cassandra.metrics.CompactionMetrics$1.value(CompactionMetrics.java:65)
	at org.apache.cassandra.metrics.CompactionMetrics$1.value(CompactionMetrics.java:57)
	at com.yammer.metrics.reporting.JmxReporter$Gauge.getValue(JmxReporter.java:63)
	at sun.reflect.GeneratedMethodAccessor31.invoke(Unknown Source)
	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.lang.reflect.Method.invoke(Method.java:497)
	at sun.reflect.misc.Trampoline.invoke(MethodUtil.java:71)
	at sun.reflect.GeneratedMethodAccessor19.invoke(Unknown Source)
	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.lang.reflect.Method.invoke(Method.java:497)
	at sun.reflect.misc.MethodUtil.invoke(MethodUtil.java:275)
	at com.sun.jmx.mbeanserver.StandardMBeanIntrospector.invokeM2(StandardMBeanIntrospector.java:112)
	at com.sun.jmx.mbeanserver.StandardMBeanIntrospector.invokeM2(StandardMBeanIntrospector.java:46)
	at com.sun.jmx.mbeanserver.MBeanIntrospector.invokeM(MBeanIntrospector.java:237)
	at com.sun.jmx.mbeanserver.PerInterface.getAttribute(PerInterface.java:83)
	at com.sun.jmx.mbeanserver.MBeanSupport.getAttribute(MBeanSupport.java:206)
	at com.sun.jmx.interceptor.DefaultMBeanServerInterceptor.getAttribute(DefaultMBeanServerInterceptor.java:647)
	at com.sun.jmx.mbeanserver.JmxMBeanServer.getAttribute(JmxMBeanServer.java:678)
	at com.sun.jmx.remote.security.MBeanServerAccessController.getAttribute(MBeanServerAccessController.java:320)
	at javax.management.remote.rmi.RMIConnectionImpl.doOperation(RMIConnectionImpl.java:1443)
	at javax.management.remote.rmi.RMIConnectionImpl.access$300(RMIConnectionImpl.java:76)
	at javax.management.remote.rmi.RMIConnectionImpl$PrivilegedOperation.run(RMIConnectionImpl.java:1307)
	at java.security.AccessController.doPrivileged(Native Method)
	at javax.management.remote.rmi.RMIConnectionImpl.doPrivilegedOperation(RMIConnectionImpl.java:1406)
	at javax.management.remote.rmi.RMIConnectionImpl.getAttribute(RMIConnectionImpl.java:637)
	at sun.reflect.GeneratedMethodAccessor21.invoke(Unknown Source)
	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.lang.reflect.Method.invoke(Method.java:497)
	at sun.rmi.server.UnicastServerRef.dispatch(UnicastServerRef.java:323)
	at sun.rmi.transport.Transport$1.run(Transport.java:200)
	at sun.rmi.transport.Transport$1.run(Transport.java:197)
	at java.security.AccessController.doPrivileged(Native Method)
	at sun.rmi.transport.Transport.serviceCall(Transport.java:196)
	at sun.rmi.transport.tcp.TCPTransport.handleMessages(TCPTransport.java:568)
	at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run0(TCPTransport.java:826)
	at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.lambda$run$241(TCPTransport.java:683)
	at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler$$Lambda$1/183345253.run(Unknown Source)
	at java.security.AccessController.doPrivileged(Native Method)
	at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run(TCPTransport.java:682)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

The lock protects from concurrent access through the two synchronized methods WrappingCompactionStrategy.getEstimatedRemainingTasks and WrappingCompactionStrategy.getNextBackgroundTask.

The lock is not just hold for a short time but for several minutes. We think that this combined with suboptimal values for memtable_flush_writers and concurrent_compactors caused our nodes to exhaust their heap and die after running incremental repair.

There also seems to be no other way to reduce the number of SSTables as a user except than waiting for the compactions to eventually finish which looks like it can take days.

Issue Links

• duplicates: CASSANDRA-10099 Improve concurrency in CompactionStrategyManager