unexpected redis warning logs on standby at end of call if 'start forwarding' used

[pphm-e]

rtpengine version the issue has been seen with

mr13.5.1.4

Used distribution and its version

Ubuntu 22.04.5 LTS

Linux kernel version used

6.8.0-1044-aws

CPU architecture issue was seen on (see uname -m)

x86_64

Expected behaviour you didn't see

when a session is terminated on the active instance we should not see warning logs for the session on the standby

Unexpected behaviour you saw

when a session is terminated on the active instance we see redis warning logs on the standby

This only happens if 'start forwarding' has been actioned on the session

Steps to reproduce the problem

have an active session that is forwarding media using the start forwarding command

tail the logs on the standby and terminate the session on the active

Additional program output to the terminal or logs illustrating the issue

WARNING: [21161181061222026122017@89.222.36.19]: [core] Failed to restore call ID '21161181061222026122017@89.222.36.19' from Redis: could not retrieve JSON data from redis
WARNING: [21161181061222026122017@89.222.36.19]: [core] Redis returned error to command 'DEL %b': READONLY You can't write against a read only replica.
WARNING: [21161181061222026122017@89.222.36.19]: [core] Redis returned error to command 'DEL %b': READONLY You can't write against a read only replica.

Anything else?

No response

[rfuchs]

Post the values of the relevant options (redis-...) from your config

[pphm-e]

On one instance the redis settings are

redis=10.0.0.1:6379/1
redis-write=10.0.0.1:6379/2
redis-expires=14400
subscribe-keyspace=1
redis-num-threads=8

on the other instance

redis=10.0.0.2:6379/2
redis-write=10.0.0.2:6379/1
redis-expires=14400
subscribe-keyspace=2
redis-num-threads=8

Redis is running locally, and set to master/slave based on keepalived notify script at the same time as active and stanby set on rtpengine

this may be a race condition? As sometimes, after spending more time monitoring logs, I see it does not happen

[rfuchs]

That's because the active/active configuration isn't really meant to work with replication. Have both Redis instances run as master, and make both rtpengine instances write to their own local Redis, and read from and subscribe to the other instance's Redis.