Health Checker

Status Bars

There are 4 status bars at the top left of a MeTL conversation. Measurements are stored for a period of 60 minutes.

1. Health Status. This displays an aggregate score, which includes latency (calculated time between the client and the server) and render time (time taken to render all stanzas to the board).
2. Attendance Status. This represents the ratio of current non-author conversation attendees to the expected maximum number of non-author conversation participants. For conversations that have a foreign relationship and can look up their expected attendance from a foreign system, the expected maximum comes from the foreign system. For conversations with no foreign relationship, the expected maximum is simply the count of all users who have ever joined the conversation.
3. Conversation Activity. This is represented as two bars: the top one is activity by the current user, and the bottom is activity by other users. This only shows activity that sends content to and from the server.

If the success rate of either server or latency are less than 1 (ie there have been one or more errors during the stored period), the background of the status bars becomes red until the success rate for the entire stored period returns to 1.

Health Check Graphs

There are a number of graphs on the Health Check page, one per measurement category. These include:

• render: time taken for the canvas to blit.
• serverResponse: time taken for a command to travel to the server and back.
• latency: calculated time between the client and the server.
• sendStanza (and any other clientSideFunc): time taken on the server to perform a function.

Within each graph there are four data sets, each divided into time slices according to a bucket size which gives a series of rolling averages.

The first three use the left-hand axis (time taken ms) to represent minimum, mean and maximum. These display as green, orange and oranger superimposed area graphs. The last is error count, which uses the right-hand axis to represent number of errors. It displays as a red area graph, superimposed over the others with a black edge.