4.3.0

• Report more information about aws operations.

4.2.0

• Allow passing in keys for statsum and sentry directly without going through taskcluster auth. This is pretty much only going to be useful for adding lib-monitor to auth.

4.1.0

• Add a monitor.timeKeeper(<name>) function that returns an object that can be used to keep time over some arbitrary steps in code.

4.0.0

• reportError() will no longer reject the promise when reporting to Sentry fails. It would require everyone to .catch() it all over the place and 99% of the time people shouldn't really care all that much about the success of the report. We now accept the promise with a value of either true or false depending on the success for those times when you really want to know. With the old way of doing things, an error from Sentry could cause a cascading series of failures.

3.1.0

• Add a patchAWS(monitor, service) function that allows for measuring timings of an aws-sdk client.

3.0.0

• reportError() now always returns a Promise that can be awaited for if you want to try to ensure that an error is reported. Most of the time this Promise can be ignored as the error will get reported anyway unless the process is shutting down.

2.3.1

Fix some issues where an error getting a sentry token caused recursive errors..

2.2.0

• A new utility to monitor any generic function or promise has been added under monitor.timer. #20
• All api calls are now monitored under a shared 'all' tag that is shared between all of the api endpoints, in addition to under their own names. #10
• Exceptions are now reported automatically with tags of the process that this is running in and the prefix that the monitor is under. Additional tags can be added as well if needed. #19

2.1.0

• If process is specified in the options when instantiating a monitor, it will automatically begin monitoring resource usage for the process. This process name will be overwritten if monitor.resources(...) is later specified and can be stopped by calling monitor.stopResourceMonitoring().

2.0.0

• Process resources (CPU, Memory) are no longer monitored by default, and instead must be turned on by calling monitor.resources(<process name>). This is to allow for different processes (e.g. web, worker, task-resolver) to be differentiated. monitor.resources returns a closing function that should be called when a process is shutting down in order to gracefully terminate. If it is not called, the interval will keep the process alive until it calls.