Hi Mark T. and Henri K., the incident is resolved.
The fix was applied at approximately 11:50 UTC, and database and share attach times in eu-central-1 have returned to normal levels.
The degradation came from excessive load on an internal metadata service in the region, which slowed the step that opens a database when a new session or connection starts. Established sessions were largely unaffected, while new connections from the web UI, MCP, and the Postgres endpoint could be slow to establish or time out.
Any query or connection that failed during the incident can be retried and should now succeed. We are continuing to monitor, and we are following up with preventative work, including alerting for this failure mode across all regions.
Apologies for the impact this has caused on your workflow, and thank you for your patience.
.png){kind=logo}
