Elasticsearch is intermittently unresponsive with the error message: "The remote server returned an error: (503) Server Unavailable."
Article ID: 100064337
Updated On:
Description
Error Message
Exception: The remote server returned an error: (503) Server Unavailable.
Cause
It has been observed that, under rare circumstances, Elasticsearch becomes unresponsive, indexing stops, and search halts.
The root cause of this is unknown. When Elasticsearch is in a hang state, it returns a 503 Server Unavailable error.
The issue goes away after restarting the Enterprise Vault Indexing Service.
Resolution
Enterprise Vault 14.5.2 has been updated with a fix. When Elasticsearch goes into a hang state for a certain duration, it can be recovered without manual intervention.
Time duration is based on the health check frequency and a configuration key.
For example, Enterprise Vault checks Elasticsearch's JVMHeap Usage every second, and if Elasticsearch does not respond to the JVMHeap Usage check 600 times, it is considered to have gone into a hang state. To recover from the hang state, Elasticsearch is restarted.
The threshold 600 is adjustable with a configuration key.
Configurable key for the Elasticsearch Health Check Exception threshold in EVIndexAdminService.exe.config file in AppSettings section:
KeyName-: ES_HEALTH_CHECK_EXCEPTION_THRESHOLD
Values range from 60 to 1800; the default value is 600.
For example:
Issue/Introduction
