Confluence Webhooks are not being delivered
Postmortem
Summary
On October 22, 2025, Confluence Cloud webhook events were not delivered to Connect apps and Forge apps that use webhooks via connectModules. This outage began after a backend deployment introduced a code change that prevented Confluence from recognizing registered webhook listeners. As a result, no webhook events were generated for approximately 20 hours, impacting apps that depend on webhooks. The issue was resolved by rolling back to a previous stable deployment. We’re taking steps to improve detection and rollback processes.
IMPACT
Between 01:12 UTC and 21:46 UTC on October 22, 2025, Confluence was unable to send out webhook events. As a result, none of the apps that rely on these events received them during this time. This includes Connect apps and Forge apps that use webhooks through connectModules.
Customers may have noticed outdated content appearing in their Marketplace apps, or new content not syncing.
ROOT CAUSE
The issue was caused by a change to Confluence backend systems. As a result, no webhook events were delivered to Connect and Forge apps that use webhooks via connectModules. This occurred because a Java class was loaded by 2 different classloaders, causing Confluence to fail to find registered webhook listeners and thus not emit events.
REMEDIAL ACTIONS PLAN & NEXT STEPS
Webhooks are a critical functionality for many apps. On this occasion, our system did not generate any errors because it was inadvertently using an incorrect version of a Java class, which was loaded by a different ClassLoader. As a result, the issue occurred silently and couldn’t be detected via existing automated error monitoring.
We are prioritizing the following improvement actions to help prevent this issue from reoccurring:
- Additional integration tests to identify problems before merging code to our main branch.
- Additional detectors and alerts to identify when webhook event volume drops below a certain threshold.
Exploring resync / lost event recovery mechanism without overloading Atlassian infrastructure
We acknowledge that the availability of Connect apps is important for our customers and ecosystem partners, and we will endeavour to provide support for these apps in line with our public roadmap for Connect end of life. You can read our announcement here: Announcing Connect End of Support: Timeline and Next Steps - Work Life by Atlassian
While this incident primarily impacted Connect, we would like to note that Forge currently includes additional internal capabilities, such as retry Atlassian app events and internal DLQ, that could assist with recovery in these types of incidents..
We highly encourage partners to migrate to Forge to utilize the latest webhook infrastructure.
We apologize to customers and ecosystem partners whose services were impacted during this incident; we are taking immediate steps to improve the platform’s performance and availability.
Thanks,
Atlassian Customer Support
Resolved
This outage did not affect Forge events. However, it did impact:
1) All Connect Apps
2) Forge apps that use webhooks via connectModules
Please note: Atlassian cannot replay or resend the missed webhook events. As a result, these events are permanently lost.
Recommended Action:
To recover, we recommend that all affected apps use their resynchronization (resync) mechanisms to fetch the latest data directly from Confluence. This will help ensure your app’s data is up to date and that any information missed during the outage is restored.
Once we complete our internal incident review process, we will publish a more detailed postmortem of what went wrong, along with steps we're taking to prevent this from happening again in the future.