Alessio Placitelli [:Dexter] Reporter
	Description • 9 years ago

The crash in [1] should be investigated: it could be either due to us failing to cancel outgoing ping requests (in |TelemetrySend._cancelOutgoingRequests|) or some ping gets submitted and races to trigger a sendtask again.

{
  "phase":"profile-before-change2",
  "conditions":[
    {
      "name":"TelemetryController: shutting down",
      "state":{
        "initialized":true,
        "initStarted":true,
        "haveDelayedInitTask":false,
        "shutdownBarrier":"Not started",
        "connectionsBarrier":"Not started",
        "sendModule":{
          "sendingEnabled":false,
          "pendingPingRequestCount":2,
          "pendingPingActivityCount":1,
          "unpersistedPingCount":0,
          "persistedPingCount":2,
          "schedulerState":{
            "shutdown":true,
            "hasSendTask":true,
            "sendsFailed":false,
            "sendTaskState":"wait on ping sends",
            "backoffDelay":60000
          }
        }
      },
      "filename":"resource://gre/modules/TelemetryController.jsm",
      "lineNumber":687,
      "stack":[
        "resource://gre/modules/TelemetryController.jsm:setupTelemetry:687",
        "resource://gre/modules/TelemetryController.jsm:Impl.observe:780",
        "resource://gre/modules/TelemetryController.jsm:this.TelemetryController<.observe:167",
        "resource://gre/components/TelemetryStartup.js:TelemetryStartup.prototype.observe:30",
        "null:null:0"
      ]
    }
  ]
}



[1] - https://crash-stats.mozilla.com/report/index/27da41a1-63c7-43dc-9687-060322150711

	Alessio Placitelli [:Dexter] Reporter
	Comment 1 • 9 years ago

This also seems related to bug 1173961.

Also, TelemetrySend [1] seems to use |task.then(clear, clear)| to clear the sending task when done. As we found out in a recent bug, this pattern proved to be ineffective. We should probably do as done in [2].

[1] - http://mxr.mozilla.org/mozilla-central/source/toolkit/components/telemetry/TelemetrySend.jsm#374
[2] - https://hg.mozilla.org/mozilla-central/annotate/e7e69cc8c07b/toolkit/components/telemetry/TelemetryStorage.jsm#l832

	Georg Fritzsche [:gfritzsche]
	Comment 2 • 9 years ago

Note that we can find the AsyncShutdown timeout crashes here:
https://crash-stats.mozilla.com/search/?product=Firefox&release_channel=nightly&async_shutdown_timeout=sendModule&version=42.0a1&_facets=signature&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#crash-reports

Only 4 related to Telemetry shutdown showed up so far since bug 1181553 landed.
Of those, only one has an active send task - so far this doesn't seem to be a big issue.

	Thomas Huelbert
	Comment 3 • 9 years ago

moving to new milestone - [40b7], P2

	Georg Fritzsche [:gfritzsche]
	Comment 4 • 9 years ago

There were no reports in the last week with an active send task on Nightly, lets keep watching this for a bit.

	Alessio Placitelli [:Dexter] Reporter
	Comment 5 • 9 years ago

We should re-check if this is still an issue and, if that's the case, its volume.

	Georg Fritzsche [:gfritzsche]
	Comment 6 • 9 years ago

There are <=5 of these on any channel.
I'm willing to call these odd outliers until further evidence arrives.