Possible API connection disruptions

We're about to change part of our API's load balancing configuration.

Although we don't expect problems (which is why we're not changing the service's status), a possible side effect would be that some TCP sessions connecting to our API would get dropped mid-session. The last time we made changes in this area, a few percent of connections were affected.

We'll be monitoring the API closely, and watching the forum for problem reports that look relevant. We'll revert the change if this problem does seem to be happening.

If you think this might be affecting you, do let us know, but please also proceed with normal troubleshooting. If you're getting an error message from our API, then this is not the problem area: if it were, your API session would be getting interrupted before we could deliver that error message. If your requests start timing out sometimes, but succeed when retried, then this could be the cause.


This immediately caused problems and has been reverted.