Circuit Breaking Exception

(Nitish Goyal) #1

I have recently migrated the cluster from 2.4 to 6.3.0 and I am getting circuit breaking exception every few hours on the nodes

Cluster details -
Data nodes : 13
Master nodes : 3
Query nodes : 3

We aren't using this cluster for searching in production as of now. Only indexing is happening on this cluster
"type" : "failed_node_exception",
"reason" : "Failed node [YMekd6kcQPydIHKnORSX6g]",
"caused_by" : {
"type" : "circuit_breaking_exception",
"reason" : "[parent] Data too large, data for [<transport_request>] would be [6608475256/6.1gb], which is larger than the limit of [6608348774/6.1gb]",
"bytes_wanted" : 6608475256,
"bytes_limit" : 6608348774

Heap size is 31 GB
I have overwritten this setting : 20%
With the same settings, indexing is working fine in 2.4.0, but failing in 6.3.0

Is this a bug in 6.3.0 version?


(Christian Dahlqvist) #2

Given the limit you have in place this seems expected. It sounds more like an issue with Elasticsearch 2.4 not triggering circuit breakers correctly. Why have you set to such a low value?

(Nitish Goyal) #3

Thanks for the explanation

I have set the value to 50% and restarted the failed nodes

Will post it here if I face this issue again


(Christian Dahlqvist) #4

Why are you overriding this in the first place? Why not use the default value?

(system) #5

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.