Describe the bug
When doing manual failover cluster failover, the partial sync is not working, and replica is doing full sync, even when only RDB is enabled.
Below are the logs from replica:
1459881:1459887:S 11 Jan 2024 16:57:22.875 * Before turning into a replica, using my own master parameters to synthesize a cached master: I may be able to synchronize with the new master with just a partial transfer. 1459881:1459887:S 11 Jan 2024 16:57:22.875 * Connecting to MASTER 172.20.0.9:6380 1459881:1459887:S 11 Jan 2024 16:57:22.875 * MASTER <-> REPLICA sync started 1459881:1459887:S 11 Jan 2024 16:57:22.875 # Cluster state changed: ok 1459881:1459887:S 11 Jan 2024 16:57:22.885 * Non blocking connect for SYNC fired the event. 1459881:1459887:S 11 Jan 2024 16:57:22.885 * Master replied to PING, replication can continue... 1459881:1459887:S 11 Jan 2024 16:57:22.885 * Trying a partial resynchronization (request 96323751e90c0c3643983fa464fbf4801e2fc303:1). 1459881:1459887:S 11 Jan 2024 16:57:22.924 * Full resync from master: df18ec4f5fdf0a9b1ce11a0766ebe7c2c38e74b7:139283664859 1459881:1459887:S 11 Jan 2024 16:57:22.924 * Discarding previously cached master state.
Following are imp settings in redis.conf:
appendonly no save 3600 1 repl-backlog-size 35000000
Redis version 6.
To perform the failover I am using Redis python package - https://pypi.org/project/redis/ redis_connection.cluster_failover(clusternode)
Comment From: javedsha
Can someone please help to troubleshoot this?
Comment From: javedsha
@oranagra not sure whom to tag for help. Can you pls help on this issue.
Comment From: oranagra
i'm assuming there's not much traffic at at that time and that it's not a backlog issue.
i see it requests a repl offset of 1, and gets 139283664859, so it looks like a bug (or a feature that wasn't yet supported).
this version is a little bit old, and i don't recall if this was already fixed in a newer one. maybe @madolson remembers / knows more?