Describe the issue: Randomly my master nodes which were healthy are going into crashloopbackoff state with the error message :
“[2025-07-02T08:20:41,587][ERROR][o.o.t.n.s.SecureNetty4Transport] [master-node-name] Exception during establishing a SSL connection: java.lang.IllegalStateException: transport not ready yet to handle incoming requests”
This breaks the cluster & i had to recreate the opensearch completely. The only additional thing I’ve added for 3.0.0 is the elasticsearch exporter deployment in order to scrape metrics.
Anybody faced the similar issues? Please let me know if we have any fix for this behaviour.
@shs_tech Could you share your elasticsearch-exporter manifest?
Did you notice the reported issue straight away or after a longer time? Is the cluster load higher when errors appear, or is it not related?
I have disabled the elastic exporter and deployed Opensearch 3.0. One more time we faced the issue. I have further increased the heap memory (OPENSEARCH_JAVA_OPTS) and it seems to be stable.
But i started getting another issue. Sometimes, the Opensearch Dashboards are performing very slow. The queries take much time & also snapshots which were getting completed within 2-3 mins earlier are now taking more than 45 mins to complete. We were seeing some rejections from fluentd as well.
When i inspect the dashboard page, could see that the Kibana_URL/internal/search/opensearch is taking time to complete the request.
Any idea/tuning we need to perform w.r.to Opensearch 3.0 ?
@pablo Even without elastic exporter i started getting the same error. With Opensearch 2.16.0, my master pod memory limits were 4Gi-5Gi with 5 replicas. With 3.0, i have given 8Gi-12Gi and JVM also updated accordingly. Still with this, 4 of my master nodes are in crashloopbackoff.