Connector Improvement: Allow setting to determine if (or how many times) a connector retries after failure
Our organization has a critical time that the database is under heavy usage and cannot be down.
Currently if the connector fails to sync, it retries unlimited times which slows down or crashes the database until the connector is paused.
I'd like a setting where it only tries 1 time to resync and then pauses itself.
It would also be helpful if I can set it to retry 24 hours later, so the sync happens at night when it won't disrupt thousands of users.
It makes absolutely no sense to keep retrying unlimited number of times. It hasn't worked 9x, why will the 10th time work? Meanwhile, we are getting calls about slowness and upset users.
-
Hi Daniel,
I agree. We are actually in the process of rolling out a change that will address some of your feedback because we recognize that while this behavior promotes uptime, it can be resource intensive.
- Going forward, we will continue to retry at the current cadence for the first 24 hours, as there is a chance that some issues may transient – transient connection issues are particularly common with databases during this period of time and there is still a good chance that the connector may recover.
- If the connector continues to fail after this period, we are planning to continue running the connector at the set sync frequency for another 48 hours – this covers weekend periods as there is still a small chance that the issue can resolve in this time.
- If the connector is continuing to fail after this, we will throttle to running the connector 1x per day.
- If the issue remains unresolved by day 14, we will auto-pause the connector – this is 14 days sooner than we currently pause connectors.
I'd love your feedback on this proposal. I think this should improve what you are mentioning.
Lyndsey, Product Manager
Please sign in to leave a comment.
Comments
1 comment