Open erikljung opened 6 months ago
We are also getting this error during a rolling deploy. We dont have very high frequency events so I dont believe we are losing events due to this, but I havent actually checked for this.
Would be great to hear some official advice here as I will likely just disable the error logging from this callback or force a hard rollout when upgrading the cluster, neither seem like great solutions tbh.
I've spent a couple of days fighting issues with stream events gone missing during rolling deploy. I created a minimal repro, to rule out any business logic issues, but I can still reproduce the error. It's possible that this configuration is sub-optimal or that I'm doing something wrong, so please advice me.
Setup
Orleans version: 7.2.6 Stream pub sub configuration: StreamPubSubType.ImplicitOnly Stream storage: MemoryGrainStorage for "PubSubStore" (if this is applicable with ImplicitOnly?) GrainDirectory: Redis DefaultCompatibilityStrategy: BackwardCompatible DefaultVersionSelectorStrategy: LatestVersion Using Postgres as grain storage/membership
Repro
Orleans.Streams.QueueCacheMissException
occurs on Silo B at the consumers end. This is the only error I get. No error occur at the producers endQuestions