fix Replication.Supervisor's strategoy and minimal durable-slot/back-pressure documentation in README #70
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Currently the supervision strategy for Replication.Supervisor is one_for_one. This is an issue for the following reasons: if Publisher crashes: We lost the current state. This means that until Postgres decides to send us all the needed Relations and Types messages again, we won't be able to decode any events from the Server. In the mid time everything would look ok but all events would get discarded. The only way to guarantee to get those back is to restart the Server. if Server crashes: The replication will restart at restart_lsn. All events from then to the LSN at which the Server crashed will get replayed. The means that the message inbox of the Publisher will become potentially inconsistent and will likely contain duplicate messages. If this is undesirable, one_for_all is required otherwise rest_for_one is fine.
|
Thank you! Just merged and pushed out a new release. Again, thank you for all the help on this. |
|
woohoo ^^ |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR fixes the supervision strategy of
Replication.Supervisor.We had an issue in production where we lost all events for multiple hours because the
Replication.Publisher crashed, restarted and discarded all events until we forced a full restart.
This is actually expect as Postgres only sends the Relations/Types/etc. messages when the replication connection
is started or on an alter on a specific table.
To fix this instead we changed the replication strategy from
one_for_onetoone_for_all.Also update of the README to explain minimally the
durable_slotandmessage_middlewareconfiguration options.