Wikimediastatus.net/User instructions

From Wikitech
Revision as of 05:15, 24 April 2022 by Quiddity (talk | contribs) (tweak; add tooltip info)

Normally, the page will display "All Systems Operational".

If something is known to be wrong, then instead there will be a brief description of the issue and who it affects (for example: "Wikipedia and other sites inaccessible for users especially in North America").

Updates are posted manually and will not be instant: they are written by the same engineers who are working on troubleshooting and resolving the problem.

The two high-level components on the page, "Reading" and "Editing", also provide information about who is affected: if there is an outage declared for only "Editing", then users who are logged out and only reading popular articles are unlikely to be affected.

Metrics

A key feature of the new status page is a set of five high-level metrics that give a look into the overall health and performance of the wiki environment. We chose a set of indicators that would show widespread issues as obvious deviations from normal, so that non-technical people could look at the graphs and say “ah, yes, something is wrong”. Automatically publishing these metrics means that users can have some idea that something is wrong for everyone, not just themselves, even before our engineers have a chance to post an update.

Each metric is described below.

Total request volume

The number of HTTP requests per second our content delivery network is servicing. A normal range for this metric is between 100,000 to 200,000, depending on the time of day. Either sharp increases or decreases from the usual value may indicate an issue (but also may not).

User-reported connectivity errors

The number of Network Error Logging reports we're receiving of connectivity trouble from the web browsers of users. A value of under 1 report/second generally does not indicate any trouble; higher numbers or sustained values above 2 may indicate disruptions with either Wikimedia's network connectivity or issues on the Internet at large.

Wiki error responses

Out of the total incoming requests, the amount per second that produced error responses. Error responses have many causes, from invalid user input to known bugs; the presence of errors does not necessarily indicate any issue. Up to a few hundred per second is not necessarily a cause for concern.

Wiki response time

The average duration (in seconds) that our application servers took to reply to uncached requests. Note that most users receive much faster responses due to caching in our content delivery network. Values of 5 seconds or less are normal.

Successful edits

The number of edits made per second. Includes both human and bot edits. Values below 10 per second may indicate performance issues.