Cycle through host before calling errback

Open arnaudmorin opened this issue 6 months ago • 1 comments

In oslo.messaging error callback (named _recoverable_error_callback), the code is raising an TimeoutError, which is preventing continuation of the retry in some situation. This quick fix switch the selection of next host (cycle) before calling the error callback. Then, on next ensure_connection, the new host is selected and we expect it to work correctly :)

See: lp-2096926

Jul 08 '25 15:07 arnaudmorin

what about merging the changes of 2 separate pr in a single one?

Jul 08 '25 16:07 auvipy