instance : 6ea (node : 6ea)
- cpu : 16core
- ram : 64GB
- jvm heap : 16GB
We have created one index and done full indexing like the below steps.
- create index (refresh interval : -1, replica : 0)
- full indexing with bulk (total index size is 1.8GB)
- check "commited" is true for all segments of created index (GET index_name/_segments)
- refresh and set refresh interval to 30s
- set replica to 5
When we do the last step(set replica to 5), there are so many logs "IndexNotFoundException: no segments" from master node.
And then we can't access master node because of those errors for long time..
As I described from step 4, We have checked commited status for all segments to prevent uncommited segments after _flush.
Could we know why this issue was happened and how to handle it?