Platform: Login Failures and Degraded Performance
Postmortem
Incident Overview
On August 28, 2026, Delinea Platform customers in the US region experienced intermittent login failures and slow performance when signing in or accessing their tenants, including error or timeout pages on the sign-in screen for some users. In some cases, retrying the login attempt allowed users to successfully access their tenant.
- Start: August 28, 2026 approximately 17:29 UTC
- End: August 28, 2026 approximately 19:19 UTC
Root Cause and Remediation
A caching related code change that was made on Aug 27, combined with an unusually large volume of background synchronization activity on Aug 28, exceeded the normal operating capacity of our caching infrastructure supporting login and authentication services. Connections to it began failing and other dependent services attempted to recover those connections repeatedly rather than cleanly, preventing the caching infrastructure from stabilizing on its own. This led to slow or failed login attempts and delays in identity-related operations across the US region.
Our engineering team rolled back the caching change and restarted the affected services, which cleared the failed connections and began restoring functionality, though performance remained degraded. The caching resource was scaled up to restore normal performance. We continued to closely monitor the service until full recovery was confirmed.
Preventative Actions
- Increase the capacity and resiliency of our caching infrastructure by distributing workloads across additional resources.
- Correct monitoring thresholds to ensure capacity pressure is detected proactively before it impacts platform performance.
- Limit connection-recovery retries to prevent recovering connections from compounding system load during a stress event.
- Apply rate limiting to large synchronization jobs to prevent any single job from negatively impacting platform-wide performance.
We sincerely apologize for the disruption this caused and the inconvenience to your operations. We are committed to preventing recurrence through the above actions.