Replica of repeatedly fails
There might be instances in which the Replica of process repeatedly fails and restarts itself. This can result from the Redis “client-output-buffer-limit” setting on the source database being configured to a relatively small value, which causes the connection to be dropped.
To identify this issue you can review the Redis log of the source database and find an entry in the log describing this issue.
To resolve this issue, follow the instructions below to reconfigure the buffer on the source database to a bigger value:
- If the source database is a Redis that is part of REP, run the following command in rladmin command-line interface (CLI) to increase the slave buffer size of the source database:
tune db < db:id | name > slave_buffer < value >
- If the source database is a Redis that is not part of REP, run the config set command through redis-CLI to increase the client output buffer size of the source database:
config set client-output-buffer-limit “slave < hard limit > < soft limit > < soft seconds >”
For additional information about this issue and configuration, refer to the following blog post: Top Redis Headaches for Devops – Replication Buffer.