Incident Report: Meta Publishing Outage — Instagram, Facebook & Threads
Severity: Major
Summary
On May 19, 2026, a Meta-wide network event prevented Buffer from publishing content to Instagram, Threads, and Facebook. During the event, Meta's systems were unable to reach the Buffer endpoints that host your media, so publishing requests could not be completed. The disruption lasted approximately 89 minutes, from 03:46 UTC to 05:15 UTC. Recovery was automatic once Meta's network connectivity was restored, and no action was required on your part. All channels returned to normal publishing performance, and no data was lost.
Customer Impact
Between 03:46 UTC and 05:15 UTC on May 19, 2026, publishing to Instagram, Threads, and Facebook through Buffer failed or was severely degraded:
Instagram: publish success rate dropped from ~94% to a low of 8.4%
Threads: error rate peaked at 72%
Facebook: error rate peaked at 80%
All content types (posts, reels, and stories) were affected. Posts scheduled during this window may have failed to publish. Because the event occurred during overnight hours (UTC), many affected posts were queued during this period. Connectivity and publishing success returned to normal by approximately 05:05 UTC, and any failed posts can be safely re-queued and published.
Root Cause
The incident was caused by a third-party network event on Meta's side, not by an issue with Buffer.
When you publish to Instagram, Threads, or Facebook through Buffer, Buffer sends Meta a request that points to your media (images and video) hosted on Buffer's servers. Meta's systems then fetch that media to complete the post. During this incident, network traffic from Meta to Buffer's media and webhook endpoints dropped to near zero, meaning Meta was unable to retrieve the content it needed to publish.
We confirmed Buffer's own systems were healthy throughout. There were no issues with our publishing infrastructure, no recent changes on our end, and no problems with our content delivery or security layer. The event was part of a broader Meta-wide disruption that also affected Meta's Messenger Platform and WhatsApp Business API, and was corroborated by external outage monitors.
Key Learnings
Our monitoring detected the issue quickly. Automated alerts fired within minutes, and our team was investigating before the first customer report came in.
The cause was outside our control. This was a Meta network event, so the outage itself was not something Buffer could have prevented.
Diagnosis took longer than we'd like. Some early investigation time was spent ruling out a Buffer-side cause before we confirmed the problem originated with Meta. We've identified clear steps to reach the correct diagnosis faster next time.
Recovery was clean. Once Meta's connectivity returned, publishing recovered on its own with no lasting impact to accounts or data.
Future Improvements
Faster diagnosis of upstream outages: We're improving our standard runbooks to quickly confirm when a publishing problem originates with Meta rather than Buffer, so we can identify third-party events sooner.
Better visibility into platform errors: We're improving how we capture the error responses Meta returns, which will let us pinpoint upstream issues immediately in future incidents.
Smarter retries for transient failures: We're exploring publishing retry logic that distinguishes temporary network issues from genuine content errors, so posts affected by short-lived outages can be automatically retried once connectivity is restored.
Proactive customer notifications: We're exploring ways to automatically alert affected customers when a third-party outage is detected, rather than waiting for reports.
