Client Loses Sync After Fork and Stops Receiving Updates

[shmulikza]

Is this a support request?
No

Describe the bug
The client does not properly handle process forking. After a fork, it loses synchronization and fails to receive updated values from the remote service.

To reproduce

1. Use the SDK within a forked process (e.g., a Celery worker).
2. After the fork, update a flag remotely.

Expected behavior
The client should continue to receive and reflect the latest flag updates after the fork.

Logs
N/A

SDK version
9.1.0

Language version, developer tools
python 3.9.*

OS/platform
Various Linux distributionsץ

Additional context
The issue appears specifically in forked environments, suggesting the client may not reinitialize or resubscribe correctly after the fork.

[keelerm84]

@shmulikza please refer to our documentation on worker-based servers like this.

You should be calling postfork after your worker has been forked. This will allow the SDK to re-establish those connections and receive updates once again.

[shmulikza]

@keelerm84 I would consider handling this on the SDK side.
That way, users wouldn’t need to deal with it directly or be aware of the underlying internals.

For example, this could be done using something like os.register_at_fork, with the SDK invoking client.postfork() automatically afterwards.