Details
Low Description
One of our nodes decided to stop listening on 9160 (netstat -l was showing nothing and telnet was reporting connection refused). Nodetool status showed no hosts down and on the offending node nodetool info gave the following:
nodetool info Token : (invoke with -T/--tokens to see all 256 tokens) ID : (removed) Gossip active : true Thrift active : true Native Transport active: false Load : 2.05 TB Generation No : 1382536528 Uptime (seconds) : 432970 Heap Memory (MB) : 8098.05 / 14131.25 Data Center : DC1 Rack : RAC2 Exceptions : 0 Key Cache : size 536854996 (bytes), capacity 536870912 (bytes), 41383646 hits, 1710831591 requests, 0.024 recent hit rate, 0 save period in seconds Row Cache : size 0 (bytes), capacity 0 (bytes), 0 hits, 0 requests, NaN recent hit rate, 0 save period in seconds
After looking at the cassandra log, I saw a bunch of the following:
ERROR [Selector-Thread-16] 2013-10-27 17:36:00,370 CustomTHsHaServer.java (line 187) Uncaught Exception:
java.lang.OutOfMemoryError: unable to create new native thread
at java.lang.Thread.start0(Native Method)
at java.lang.Thread.start(Thread.java:691)
at java.util.concurrent.ThreadPoolExecutor.addWorker(ThreadPoolExecutor.java:949)
at java.util.concurrent.ThreadPoolExecutor.execute(ThreadPoolExecutor.java:1371)
at org.apache.cassandra.concurrent.DebuggableThreadPoolExecutor.execute(DebuggableThreadPoolExecutor.java:145)
at org.apache.cassandra.thrift.CustomTHsHaServer.requestInvoke(CustomTHsHaServer.java:337)
at org.apache.cassandra.thrift.CustomTHsHaServer$SelectorThread.handleRead(CustomTHsHaServer.java:281)
at org.apache.cassandra.thrift.CustomTHsHaServer$SelectorThread.select(CustomTHsHaServer.java:224)
at org.apache.cassandra.thrift.CustomTHsHaServer$SelectorThread.run(CustomTHsHaServer.java:182)
ERROR [Selector-Thread-7] 2013-10-27 17:36:00,370 CustomTHsHaServer.java (line 187) Uncaught Exception:
java.lang.OutOfMemoryError: unable to create new native thread
at java.lang.Thread.start0(Native Method)
at java.lang.Thread.start(Thread.java:691)
at java.util.concurrent.ThreadPoolExecutor.addWorker(ThreadPoolExecutor.java:949)
at java.util.concurrent.ThreadPoolExecutor.execute(ThreadPoolExecutor.java:1371)
at org.apache.cassandra.concurrent.DebuggableThreadPoolExecutor.execute(DebuggableThreadPoolExecutor.java:145)
at org.apache.cassandra.thrift.CustomTHsHaServer.requestInvoke(CustomTHsHaServer.java:337)
at org.apache.cassandra.thrift.CustomTHsHaServer$SelectorThread.handleRead(CustomTHsHaServer.java:281)
at org.apache.cassandra.thrift.CustomTHsHaServer$SelectorThread.select(CustomTHsHaServer.java:224)
at org.apache.cassandra.thrift.CustomTHsHaServer$SelectorThread.run(CustomTHsHaServer.java:182)
There wasn't anything else overtly suspicious in the logs except for the occasional
ERROR [Selector-Thread-0] 2013-10-27 17:35:58,662 TNonblockingServer.java (line 468) Read an invalid frame size of 0. Are you using TFramedTransport on the client side?
but that periodically comes up - I have looked into it before but it has never seemed to have any serious impact.
This ticket is not about why an OutOfMemoryError occurred - which is bad but I don't think I have enough information to reproduce or speculate on a cause. This ticket is about the fact that an OutOfMemoryError occurred and nodetool info was reporting Thrift active : true and Exceptions : 0.
Our monitoring systems and investigation processes are both starting to rely on on the exception count. The fact that it was not accurate here is disconcerting.