@unhuman -
It’s hard to get a complete picture of what’s going on here, but here are my initial thoughts:
1 - Not sure what exactly caused this specific timeout, but they can happen for a number of reasons - It could be you overloaded your server hardware or a network glitch. The client will timeout an operation before it blocks indefinitely.
2 - I suspect the garbage is because the operation was terminated mid-flight, so you have a partial read.
3 - I am not 100% sure why this would occur tbh - it could be a side effect of a timeout.
One thing to note with muxio is that you have multiple operations being read at that same time or waiting to write, if a connection is torn down all operations using that connection will fail. Also, the 2.4.0 release has several bug fixes and improvements for muxio and 2.4.1 will also.
If you get to a point where this is reproducible, please create a Jira ticket. Additionally, I’ll have our QE team monitor this while we test.
Thanks!
-Jeff