This is a standing note about how we communicate work on the platform, not a report of anything that has happened. It is here so that you know where to look, and what it means when there is nothing to find.
Where notices are posted
This announcements page is the record. The dashboard in your client area reads the same list, so you will see current notices when you sign in without having to come here. An announcement that concerns an incident also appears on the status page, alongside the live state of the machines we monitor.
Planned work
Work that is scheduled in advance and that could interrupt a service is posted here before it starts. The notice says what is affected, when the window opens, and roughly how long we expect it to take. Routine work with no customer-visible effect (a package update, a configuration change applied without restarting anything) is not announced, because a page full of notices about nothing is a page nobody reads.
Emergency work
Some work cannot wait for a notice: a security patch that is being actively exploited, hardware that is about to fail, a provider carrying out work at short notice on their own network. In those cases the work is done first and written up here afterwards, with what happened and what we changed.
What we do not do
We do not send a separate e-mail for every piece of routine work. We do not post notices to pad this page out, and we would rather it stay short.
If something is wrong and there is no notice
Then it is probably specific to your service rather than to the platform, and the fastest route is a ticket from your client area. Include the service it concerns and what you are seeing. A ticket that arrives attached to the service is one that can be looked at directly instead of identified first.