Resolved incident
Login errors in US region
Keeper · Major · Started
- Duration
- 19m
- Severity
- Major
- Status
- Postmortem
Summary
Major · Postmortem
At 9:21AM PST we received notifications of API errors coming from our backend systems. The issue resolved itself without any required action at 9:31AM PST. During that 10 minute period, there was a database lock in the primary writer instance causing new login sessions to fail. Recently, Keeper upgraded all production RDS Aurora databases to MySQL 8.x. This updated database engine is sensitive to one particular SQL query that was causing large table scans to occur. The code changes and QA process was already under way when this issue occurred. The backend code change which addresses this issue passed QA at 12:48PM PST and we released the fix to prevent this issue from occurring again \(Release Jira ticket REL-4633\). We’re keeping a close eye and will issue additional updates as necessary.
Event timeline
How this incident unfolded, from the official status page.
Identified
Keeper
We are aware of a high number of errors in the US region. The issue will be fixed shortly.
Resolved
Keeper
We have resolved the issue and plan to deploy a backend update shortly.
Postmortem
Keeper
At 9:21AM PST we received notifications of API errors coming from our backend systems. The issue resolved itself without any required action at 9:31AM PST. During that 10 minute period, there was a database lock in the primary writer instance causing new login sessions to fail. Recently, Keeper upgraded all production RDS Aurora databases to MySQL 8.x. This updated database engine is sensitive to one particular SQL query that was causing large table scans to occur. The code changes and QA process was already under way when this issue occurred. The backend code change which addresses this issue passed QA at 12:48PM PST and we released the fix to prevent this issue from occurring again \(Release Jira ticket REL-4633\). We’re keeping a close eye and will issue additional updates as necessary.