Partial Outage

Summary

On May 31, 2026, some MyCashflow stores were unavailable after one of our database servers stopped working correctly during a routine automated maintenance operation. The server remained switched on and reachable, so monitoring did not recognize that anything was wrong. As a result, the problem went unnoticed for several hours before it was found and fixed manually.

Impact

A subset of merchants were affected. Their stores and admin panels were unavailable during the incident. All other merchants were unaffected.

Outage window: May 31, 2026, 00:05–08:45 EET.

Root Cause

A core service on one of our database servers stopped during a routine automated maintenance operation and did not restart on its own.

Detection

The affected server stayed switched on and remained reachable the whole time, so automated checks continued to report it as healthy. The failure was with the services running on the server rather than with the server itself, which the monitoring was not set up to catch. The issue was identified only after manual investigation.

Resolution

Once the problem was identified, the affected service was restarted and the affected stores came back online.

What we're doing going forward

Server monitoring is handled by our data center partner. In this case, their monitoring rules did not catch a failure that occurred within the services running on the server, even though the server itself was healthy. Together with our data center partner, we are improving their monitoring system to add several layers of monitoring so that a problem is caught no matter where it occurs:

• Direct monitoring of the core services on each database server, so an internal failure raises an alert immediately — even when the server itself appears healthy.
• Direct health checks on the databases themselves, to confirm they are responding correctly.
• External checks that load a test store the same way a customer would, confirming the whole service is working end to end.

Together, these ensure that future problems are detected quickly and automatically.