[Bucardo-general] Should Bucardo Treat Deadlocks as Serialization Failures?

[David E. Wheeler]

Fellow Bucardoans,

I’ve just committed a test for serialization failures, to make sure that Bucardo properly retries after the failure. It’s working great.

Kevin Grittner tells me via email that he considers deadlocks to also be serialization failures. Currently, it looks like Bucardo errors out on a deadlock failure. I’m wondering if we shouldn’t change it to use exactly the same model as a serialization failure: sleep and try again. Thoughts?

Best,

David