May 5, 2026 Outage - 7:15 AM UTC
Drexel
· May 10, 2026
An outage started on PluralSpace at 7:15 AM UTC.
Timeline
- 6:45 AM UTC: An update was pushed to production
- 7:15 AM UTC: Errors started increasing on PluralSpace
- 7:17 AM UTC: Alerts sent to developer
- 7:20 AM UTC: Investigation of the errors started
- 7:25 AM UTC: Initial fix attempted
- 7:30 AM UTC: Secondary fix implemented
- 7:34 AM UTC: Sites back online, fixes applied fully.
What was it?
The update pushed to production included a new column on a few tables, that had a default value. The migrations applied in ~5 seconds, but the connection pool's query cache didn't properly clear.
This resulted in a 500 error on the site, and a cached plan must not be changed error in the database.
Restarting the servers (and the corresponding connections) didn't clear the cache, the stale caches persisted.
The connection pool was recreated and applied, which cleared the query cache.
Was this helpful?