Circuit breaker exception Elasticsearch

[cool_guy99 (Tirth Mehta)]

Im seeing the following errors while creating the index

failed shard on node : failed recovery, failure RecoveryFailedException[[index_1][1]: Recovery failed from {Node_name}{Node_name}{Node_name}{NodeIP}{NodeIP:9300}{rack_cloud=Cloud1} into {es-data-Node}{Node_name}{Node_name}{NodeIP}{NodeIP:9300}{rack_cloud=Cloud1}]; nested: RemoteTransportException[[Node_name][NopeIP:9300][internal:index/shard/recovery/start_recovery]]; nested: CircuitBreakingException[[parent] Data too large, data for [<transport_request>] would be [20412311022/19gb], which is larger than the limit of [20293386240/18.8gb], real usage: [20412250920/19gb], new bytes reserved: [60102/58.6kb]];

This happened a few days ago, but when I tried to check the node stats currently, I do not see any higher parent circuit breaker limit reached in any of my nodes. How can I clear my unassigned shards here?

[HenningAndersen]

Hi @cool_guy99,

the log file looks like it is taken from a 7.0+ system. Are you using G1 GC? If so, setting

-XX:G1ReservePercent=25
-XX:InitiatingHeapOccupancyPercent=30

might help here and I would be very interested in hearing the results.