Monday October 2nd 2023, 11:51 UTC - Wednesday October 4th 15:20 UTC
A small percentage of customer Appium tests failed when certain commands were executed, like capturing screenshots or executing custom scripts.
A connection pool used by our Appium server was intermittently exhausted during spikes in the usage of mid-session install scripts. During these spikes, we were leaking HTTP client connections, which caused timeouts and eventual connection issues to the Appium server.
We restored the functionality of the service in two ways:
During the incident, we addressed the HTTP client connection leak and enhanced monitoring to better identify this specific issue should it happen again in the future.