https://build.chromium.org/p/tryserver.chromium.perf/builders/android_fyi_perf_bisect/builds/1306

This is running again.


From what I can see, these are actually cron jobs that run once a day and kick off a bunch of preset bisects (haven't actually downloaded the configs yet but they're stored as a MultipartEntity, key: 'bisect_fyi_config_map'). Digging through the logs I see seemingly harmless failure setting the bisect status on the bug since the bug doesn't exist in the datastore (I can just add it). The failure seems to come from checking whether the bug was updated (it was, https://bugs.chromium.org/p/chromium/issues/detail?id=577371#c1382).

There are also some that seem to have been failing to update  for months.
https://bugs.chromium.org/p/chromium/issues/detail?id=584455

Will poke around with some additional logging tomorrow.

Got another alert, this time about a result mismatch:

Bisect result is not as expected: Invalid type. Expected: . Actual: .. 

A Bisect FYI Try Job for positive_culprit_mac was submitted on mac_fyi_perf_bisect at http://build.chromium.org/p/tryserver.chromium.perf/builders/mac_fyi_perf_bisect/builds/978.
Bot:	mac_fyi_perf_bisect
Test Case:	positive_culprit_mac
Bisect Config:	
config = {
  "bad_revision": "358925",
  "bug_id": 567233,
  "command": "src/tools/perf/run_benchmark -v --browser=release --output-format=chartjson --also-run-disabled-tests dummy_benchmark.stable_benchmark_1",
  "good_revision": "358915",
  "gs_bucket": "chrome-perf",
  "max_time_minutes": "20",
  "metric": "gaussian-value/dummy_page.html",
  "recipe_tester_name": "mac_fyi_perf_bisect",
  "repeat_count": "20",
  "test_type": "perf"
}
Error Details:	
None
Bisect Results:	
None