fix(duration accuracy for coldstarts by using process.uptime) #292
Conversation
Signed-off-by: Erica Windisch <73207+ewindisch@users.noreply.github.com>
ewindisch
changed the title
|
@ewindisch re-ran the workflow and tests are green now -- if your commit is formatted in the semantic release formatting (you can do this by running |
There was a problem hiding this comment.
According to the node docs:
https://nodejs.org/api/process.html#process_process_uptime
process.uptime() is the time since the process was started, the duration is meant to measure invocation time. Since a process can encompass many invocations and the process gets frozen in periods of inactivity, I'm not sure this accomplishes the goal of making the duration more accurate. Maybe there's a way to store the uptime before the process is frozen ad then compare the current uptime against that value for each invocation?
| @@ -212,6 +212,7 @@ class Report { | |||
|
|
|||
| if (this.report.coldstart) { | |||
| this.report.labels.add('@iopipe/coldstart'); | |||
| this.report.duration = Math.ceil(process.uptime() * 1e9); | |||
There was a problem hiding this comment.
@ewindisch Isn't this the uptime since the process was created? Wouldn't that encompass many invocations?
|
Closing as stale |
Fixes #283
Signed-off-by: Erica Windisch 73207+ewindisch@users.noreply.github.com