We had a 1 hour window where a small percentage of requests failed intermittently across multiple different servers all with RedisCluster::get(): SSL: Connection reset by peer when trying to connect to a Serverless Elasticache instance.
Nothing in the Elasticache instance's metrics is out of the ordinary nor shows any sign of load. I can see a handful of new connections established over the hour, it roughly correlates with the errors, and I think just represents the connections getting re-established on a future request after the failure.
I'm using PhpRedis's RedisCluster as my client, and have configured redis.clusters.persistent=1. This article would imply that PhpRedis should be retrying under the hood, so i'm surprised that wouldn't prevent failures.
Is there some trick to maintainting stable/robust connectivity to Elasticache? I have Laravel configured to store session there, so it's critical that we be able to access it reliably. (it seems common to use redis for this, but maybe not elasticache? is there a more reliable setup?)
