Edge 14 seems to get stucks 25% of the time

[workmanw]

We've been noticing more and more that Edge 14 seems to be getting "stuck" when running tests. (FWIW we're using browserstack via testswarm-browserstack). After a bit of debugging, I'm seeing this happen on about 25% of the Edge 14 runs:

"Edge": {
  "data": {
    "browserFamily": "Edge",
    "browserMajor": "14",
    "browserMinor": "",
    "browserPatch": "",
    "osFamily": "",
    "osMajor": "",
    "osMinor": "",
    "osPatch": "",
    "deviceFamily": "",
    "deviceMajor": "",
    "deviceMinor": "",
    "displayInfo": {
      "class": "swarm-icon swarm-browser swarm-browser-edge swarm-browser-edge-14",
      "title": "Edge 14",
      "labelText": "Edge 14",
      "labelHtml": "Edge 14"
    }
  },
  "stats": {
    "onlineClients": 1,
    "activeRuns": 0,
    "pendingRuns": 1,
    "pendingReRuns": 0
  }
}

For whatever reason we have an onlineClient and a pendingRuns, but it's just sitting there not executing the job. I'm curious if anyone else has seen this problem?

EDIT: Also worth noting that when I watch the automated runs, I can see that all the tests pass, the runner goes into "cooling down mode", then cycles back to what looks like "polling mode". But it didn't report the successful test run, and it doesn't seem to pick up the test for a rerun.

Maybe the runner knows it just ran that specific test? IDK. But it definitely seems like when it gets into this funk state, it's like it failed to report the successful run back to testswarm.

[mgol]

We're seeing a related issue in jQuery Core tests: http://swarm.jquery.org/project/jquery. Edge 14 times out on AJAX tests all the time; the other ones pass. Note that on Edge 13 they all mostly pass.

[workmanw]

@mgol Man that's good to know. I thought I was going crazy. I have some cycles I can dedicate to this, but I don't really know where to begin. Do you have any thoughts?

[jcook793]

@workmanw you still could be crazy

[Krinkle]

Closing for now. I believe the issue was a combination of Edge 14's user-agent being slightly off, ua-parser not detecting it as such, browserstack running a point-version of it, and testswarm-browserstack not mapping it.

Given it seems to have resolved it self, closing this issue for now.

[workmanw]

I'm not sure I agree with that assessment, but it's not a big deal.