Catch TimeoutError

[bokolob]

Some network problems could cause TimeoutError, but it isn't handled. Moreover, the thread dies silently without any attempts to recover.

The bug can be also reproduced by setting the socket timeout option in Redis URL.

Probably there are other exceptions that should be retried as well, e.g BusyLoadingError.

[miguelgrinberg]

Any idea what's the equivalent error for aioredis so that both are fixed in parity?

[bokolob]

No idea yet, I don't use async server :( It's unclear from the first glance and needs some investigation.

[bokolob]

By the way - It's very upsetting when thread dies, and the rest of the application continues to live. I'd fix that - register a callback to call if thread stop, or crash the whole app (in flask-socketio case)

[roymiloh]

Hi @miguelgrinberg! I’m noticing an increase in the number of Redis connections (I’d expect a single connection per instance), and I suspect it might be related to this issue. I saw your commit - since RedisError isn’t necessarily a connection issue, it seems possible that new connections keep getting opened while the old ones remain active. Before I work on a fix, does it make sense to you?