During Valkey/Redis upgrade in AWS ElastiCache cluster all IPs of master servers get changed(Swapped with slaves or completely new - We don't know yet). If Slots caching is enabled(And only AWS DNS name is passed to the class constructor and used as a hash key in caching) it uses the old IPs until PHP_FPM is restarted to reset the cache, causing all operations in a following requests to fail with an error "Can't communicate with any node in the cluster".
PR:
#2807
We didn't test the codechange yet, until next cluster upgrade.
During Valkey/Redis upgrade in AWS ElastiCache cluster all IPs of master servers get changed(Swapped with slaves or completely new - We don't know yet). If Slots caching is enabled(And only AWS DNS name is passed to the class constructor and used as a hash key in caching) it uses the old IPs until PHP_FPM is restarted to reset the cache, causing all operations in a following requests to fail with an error "Can't communicate with any node in the cluster".
PR:
#2807
We didn't test the codechange yet, until next cluster upgrade.