At 7:59 UTC we were informed by our monitoring systems that database performance changed significantly. During investigation we found out, that problems occurred after performing a routine database migration. The WFM service was either rebooting or starting with a previous release. Only a small amount of customers were effected due to our upgrade mechanism.
At 09:28 UTC we applied a fix for the effected database migration. The WFM service operates now as expected again.
We enhanced our monitoring and fallback mechanism to avoid such a situation in future.
We apologise for the service disruption.