Routine maintenance is in progress. The site will be available again within the hour, with no action required from your side.
We're rotating the storage controller in cabinet B-12 and applying a kernel upgrade across the worker pool. The job is on plan and on schedule.
The public website and customer portal return this page until the maintenance ends. Inbound webhooks pause and resend automatically; queued jobs replay; in-flight sessions reconnect transparently once we're back.
Behind the scenes, this upgrade brings:
By 15:40 UTC. If we're still showing this page at 15:55, that's a sign something didn't go as planned — please write to ops and we'll respond personally.
Anything you submitted up to one minute before the window is committed to durable storage and replicated to followers. Nothing in the last 60 seconds before the window is lost — we wait for replication to settle before entering maintenance mode.
No. Sessions are stored in a separate identity service that is not affected by this maintenance.
Roughly once a quarter for storage and kernel upgrades. We try to schedule them at the same hour of the same day of the week so customers can plan around the predictability.