For whom: panel administrator, on-duty engineer, support. Required rights:
statuspage.view for reading, further on by actions (see Rights).Why is it a separate server?
The status page doesn’t go where the panel is, and that’s the main solution in its device. The panel is something that customers can’t see and can lie with the rest of the infrastructure. The page has to respond when everything else is not available, so it has its own server, its own database and its own certificate. Here are some things that are useful to know in advance:- The panel doesn’t store observations. It doesn’t store uptime history, it doesn’t store checks, it doesn’t store “last statuses,” it lives on the server of the page. It tells you what to show, not what happened.
- The page performs its own checks, and if the panel is turned off, the HTTP and TCP checks will continue, and the visitor will see the actual picture.
- The state of the nodes and the Flow only knows the bar, so it sends it to the page once a minute. When the panel is silent longer than it should be, the component is labeled obsolete, rather than staying with the last good value.
- You can’t remove the page server from the dashboard. The deletion button erases the settings in the dashboard; the machine on which the page is standing keeps working.
Sections
A tab that you do not have the right to do is not shown.
Conditions that the visitor sees
The overall status of the page is collected from the component states and active incidents:unknown means exactly what it says: no data. It tags a component whose source has not been answered in a long time, and it looks like a newly created page that has nothing to show yet.
How edits get to the page
There’s no separate publishing step. The saved edit goes to the page itself, a fraction of a second after saving; while it’s being sent, the section header burns, “Send to the page…” If the page is not available at this point, the panel will tell you about it there and repeat it later. The drafts are left where they’re really needed, at the incident and at the service window, and that’s the property of the record itself: the draft is only visible in the panel, the published recording is visible to everyone.Preservation, publication and runtime
The dashboard stores the changes, the runtime page serves the visitors. A saved component or incident does not mean a successful delivery of a new image. When runtime is not available, the changes remain waiting and are re-delivered after the connection is restored. Retrying with the same content uses a pending revision; you don’t need to change the text for a new number. Check the result of the publication and the actual public page. The running process or HTTP 200 from another site is not confirmation of Statuspage’s readiness.Rights.
The rights are granted to the account in the Access section and the session and are distributed by the actions, not by the all-or-nothing principle. | Right | What the | Opens |---|---| |statuspage.view | Enter and read everything, including the sync state of |
| statuspage.edit | Design, groups, components, checks, access mode |
| statuspage.incidents | Incidents and records in their chronology |
| statuspage.maintenance | Planned operations and transitions between their states |
| statuspage.deploy | Installation, reinstallation, domain, access keys, removal of |
On duty, which leads incidents, enough statuspage.view and statuspage.incidents: to touch the composition of components or demolish the installation, he will not be able.
Linkage to support
The support module reads the status of the page and uses it in two places: the ticket card shows what is now with the services, and the AI response relies on the same thing. If the status data is outdated, it is pronounced aloud - “last known”, not “current”. A message that turns out to be a general failure rises from support to an incident page, and a draft is created: a person publishes it. Incidents and planned work.Restrictions
- You can see one page in the section that’s set, and it makes sense to have more than one page infrequently, and the number of pages is limited by the tariff.
- The page does not keep its schedule of repetitive works (every Tuesday, the first day of the month): the windows wind up one at a time.
- The metrics on the graphs are collected by the page itself. Uptime history cannot be transferred from another status service, it begins from the moment of installation.
