Exception cause unwrapping ran for 10 levels

[jasonwee]

Hi,

To summarize from https://groups.google.com/forum/#!msg/elasticsearch/WdMU4_vD8zk/SKjOiCGHvFgJ

running elasticsearch version 0.90.7

gist file
https://gist.github.com/jasonwee/8282477
https://gist.github.com/jasonwee/8294514

server and client both running the same version the entire times.
$ java -version
java version "1.6.0_25"
Java(TM) SE Runtime Environment (build 1.6.0_25-b06)
Java HotSpot(TM) 64-Bit Server VM (build 20.0-b11, mixed mode)

According to Jörg Prante, this is worth raising an issue at the Elasticsearch github .

node3 elasticsearch instance was restarted and then the exceptions is no longer exists in node1 and node4 no more.

If you need more information, please let me know what I can provide.

Thank you.

Jason

[clintongormley]

Hi @jasonwee

Sorry it has taken so long to look at this. That's the first time I've seen this particular exception. Have you ever seen this problem again? The code has obviously changed a lot since back in 0.90.7. I think it is not worth investigating at this stage, but if you see something similar on a more recent version, please open another ticket.

thanks

[jasonwee]

No, I have no seen this exception anymore. Thank you for this response.

However, this one happened recently. Twice actually. https://groups.google.com/forum/#!topic/elasticsearch/1TlJDwuKXiA

Should I file a git issue?

[clintongormley]

Hi @jasonwee

That's on 0.90.7. I suggest upgrading as we're no longer supporting the 0.90 branch

[jasonwee]

sure sure, upgrade is good, could you please tell me which issue that fix the exception?

[sacheendra]

This problem still exists. Noticed it while doing a benchmark. Some nodes had high CPU usage relative to other nodes and similar things were there in the logs of such nodes.

Gist of the error: https://gist.github.com/sacheendra/ab00b6625c3367ed06c1

ES version: 1.4.0
JVM: Hotspot 1.7.0_67
Number of nodes: 15

[Srinathc]

Found a similar exception on release 1.3.0.
https://discuss.elastic.co/t/stackoverflowerror-in-es/25278/2

[clintongormley]

We have changed exception handling completely in master - would be interested to know if you're seeing something similar in 2.0 (once it is released)

[Srinathc]

Are there any details on when and why this happens? Any remedies for 1.x releases will be great to know. Our instances are already on production.

Thanks.

[clintongormley]

almost all of these exceptions are due to a change in serialization between java versions. we no longer rely on java serialization. also, stack traces are now trimmed of their guice content (see #13782)

I think this is resolved in 2.0