Status
These are counts from the audit tool itself, read when you open this page. They are approximate, and this page is not an uptime promise. There is no green banner here on purpose, because a banner I typed by hand would say nothing about whether an audit works right now.
Recent audits
The counts load with a small script, and it has not run in this browser. The same numbers are at /api/status as plain data.
Incidents
The incident list loads with the same script. Without it, the list is at /status.json as plain data.
How these are counted
- Every audit that starts adds one to the started count. An audit that returns a report adds one to the reports returned count.
- An audit that does not finish is counted once, under the reason it stopped. A site that blocks the crawler in its robots.txt, answers with a refusal, or cannot be reached counts as the site refusing or being unreachable. Those are not faults in the audit tool, and the crawler page explains how to let it in.
- The rate limit is protection against someone hammering the tool, and it counts on its own row.
- An audit that did not finish and has no recorded reason is counted as failed on our side, so that row leans towards blaming the tool.
- Audits from my own browser are left out when I have that switch on. Under heavy load a count can be missed, so treat every number as approximate.
- The percentages are shares of audits started. With only a handful of audits a day, one failure can look like a large share, which is why the counts sit beside the percentages.
The incident list
I keep the incident list by hand, in a file called status.json. Each entry has a date, what happened, who it affected and what I did about it. It only holds what I noticed and chose to write down, so an empty list means none that I know of, not a guarantee.