[Serverless Containers] Scaling issues on Private Network

After further investigation and gathering more metrics, we noticed that the errors seen last week when reopening the status are not related to this incident. We will close it now, sorry again for any inconvenience.

Despite the fix deployed last wednesday, we are still seeing a few functions/containers instances (using PN integration) unable to start; even if less occurrences happen now.

We are investigating, sorry for any inconvenience.

A fix for the race condition in private network management has been deployed across all regions. We are now closely monitoring the systems to ensure full resolution.

Initial reports indicate that the issue is resolved, but we continue to observe for any residual effects. Thank you for your patience, and we apologize for any disruption this may have caused.

Unlike said in the initial message, note that this race condition, despite rare, can be triggered for functions and containers on all regions.
We have also seen rare occurrences of this bug in the previous weeks.

The root cause has been identified. Due to a race condition in the private network management, it can occasionally happen that some functions/containers instances are not able to start.

This race condition stays rare, and we are working on a fix. In the meantime, in case a function/container attached to a PN cannot scale up, deploying again the function/container can solve the issue.

Sorry for any inconvenience.

Since 2025/11/14, some Serverless Functions and Containers attached to a private network in the fr-par region may experience scaling issues.
This can result in request timeouts for functions/containers scaling from zero, or failure to scale up for those with a minimum scale of 1. We are currently investigating this issue, which only affects functions/containers attached to a private network.