Troubleshoot zookeeper connection failures #310
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
and I'm guessing it could be due to throttling
Kafka for example reports:
kafka.zookeeper.ZooKeeperClientTimeoutException: Timed out waiting for connection while in state: CONNECTING
And zookeeper logs include things like:
java.io.EOFException
at java.base/java.io.DataInputStream.readInt(DataInputStream.java:397)
at org.apache.zookeeper.server.quorum.QuorumCnxManager$RecvWorker.run(QuorumCnxManager.java:1206)
[2020-03-06 12:35:45,056] WARN Connection broken for id 1, my id = 4, error = (org.apache.zookeeper.server.quorum.QuorumCnxManager)
java.net.SocketException: Socket closed
at java.base/java.net.SocketInputStream.socketRead0(Native Method)
at java.base/java.net.SocketInputStream.socketRead(SocketInputStream.java:115)
at java.base/java.net.SocketInputStream.read(SocketInputStream.java:168)
at java.base/java.net.SocketInputStream.read(SocketInputStream.java:140)
at java.base/java.io.BufferedInputStream.fill(BufferedInputStream.java:252)
at java.base/java.io.BufferedInputStream.read(BufferedInputStream.java:271)
at java.base/java.io.DataInputStream.readInt(DataInputStream.java:392)
at org.apache.zookeeper.server.quorum.QuorumCnxManager$RecvWorker.run(QuorumCnxManager.java:1206)
I've been able to resolve it twice with this memory limit increase,
but of course it could also be that the restart resolves the issue.
solsson
added a commit
that referenced
this pull request
Apr 23, 2020
through the service. This helps troubleshoot issues like #310 by pointing out (by podIP) which actual zookeeper connection that failed. Also I like the simplification.
|
The issue is present in clusters with preemptive notes as well. I just found https://issues.apache.org/jira/browse/ZOOKEEPER-2938 with fix https://issues.apache.org/jira/browse/ZOOKEEPER-2164. I just found those issues so on next failure I'll try to only delete the leader. Until now I've scaled down and up again. |
|
https://zookeeper.apache.org/doc/r3.6.1/releasenotes.html contains:
They might not be the cause for timeouts, but definitely pose difficulties for stabilization on ephemeral/shared-core nodes. |
|
After testing both Zookeeper 3.5.7 and 3.6.1 with a lot of different configs the essence of the failure remains. A kafka pod fails to start up, with timeouts like: The amount of retries depends on the combination of
And Kafka 2.5.0 increases the default session timeout to 18s based on a reasoning around transient instability vs genuine failures. I haven't found any effect of varying these timeouts on the actual cause of the crashloop. |
solsson
added a commit
that referenced
this pull request
Jul 12, 2020
solsson
added a commit
to solsson/dockerfiles
that referenced
this pull request
Aug 16, 2020
Yolean/kubernetes-kafka#310 because when I investigated the timeouts there I found reports that similar errors had been fixed by cleaning up (or trashing) persisted state. > Detected a direct/mapped ByteBuffer in the image heap. > A direct ByteBuffer has a pointer to unmanaged C memory, > and C memory from the image generator is not available at image run time. > A mapped ByteBuffer references a file descriptor, > which is no longer open and mapped at run time. Error: com.oracle.graal.pointsto.constraints.UnsupportedFeatureException: Detected a direct/mapped ByteBuffer in the image heap. A direct ByteBuffer has a pointer to unmanaged C memory, and C memory from the image generator is not available at image run time.A mapped ByteBuffer references a file descriptor, which is no longer open and mapped at run time. To see how this object got instantiated use -H:+TraceClassInitialization. The object was probably created by a class initializer and is reachable from a static field. You can request class initialization at image run time by using the option --initialize-at-run-time=<class-name>. Or you can write your own initialization methods and call them explicitly from your main entry point. Detailed message: Trace: at parsing org.apache.zookeeper.server.persistence.FilePadding.padFile(FilePadding.java:78) Call path from entry point to org.apache.zookeeper.server.persistence.FilePadding.padFile(FileChannel): at org.apache.zookeeper.server.persistence.FilePadding.padFile(FilePadding.java:76)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I've experimented with a 3-kafka 5-zk cluster on a 3-node GKE e2-small cluster, i.e. just enough resources to run the Kafka stack together with some basic infra. During tests I occasionally get kafka pods restarting due to failures to connect to zookeeper. Also clients like the kafka-topics CLI had troubles connecting. No zookeeper restarts, no OOMKilled. See the first commit comment for some stack traces.