When a bus process restarts, every client on every database it serves loses its websocket at the same instant and immediately schedules a reconnection. Each database that comes back triggers a registry recompute server-side, so a tight reconnection window makes all of those recomputes land together and pile up on the freshly started process.

The jitter added to each reconnection delay was only one second, far too narrow to break up that wave. Widen it to thirty seconds so the retries of the many clients fan out over a much wider window and the registry recomputes spread over time instead of colliding.

Raise the ceiling on the retry delay to two minutes to match that wider spread, and drop the exponential growth factor: with a thirty-second jitter accumulating on every attempt, the delay already climbs on its own, so scaling it further only pushed clients toward the ceiling sooner without spreading them any better.

Description of the issue/feature this PR addresses:

Current behavior before PR:

Desired behavior after PR is merged:




---
I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr

Forward-Port-Of: odoo/odoo#278075
Forward-Port-Of: odoo/odoo#276869