Not sure if the Mac part of this report is related to the ongoing work on the Mac bots, looping in ellyjones@ and kbr@

I recently landed <https://chromium-review.googlesource.com/c/chromium/src/+/1025638> for Mac - the timeouts could (?) be related to  issue 828031  overall, but I'm not sure if/how that CL could have introduced bot timeouts.

The Mac failures are caused by  crbug.com/828031 . I'll look into the Linux red.

we should be looking at why those shards are getting picked up as timeouts and not failures. swarming certainly doesn't seem to think that they're timeouts, e.g. https://chromium-swarm.appspot.com/task?id=3d152f8e997ea710&refresh=10&show_raw=1

Re #4, I think these shards are identified as failures, am I understanding it incorrectly?

browser_tests on Mac-10.11 Run on OS: 'Mac-10.11'
Max pending time: 0:01:02.471440 (shard #9)
Max shard duration: 34s (shard #7)
stdout
some shards did not complete: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9
swarming.summary
step_metadata
shard #0 (failed)
shard #1 (failed)
shard #2 (failed)
shard #3 (failed)
shard #4 (failed)
shard #5 (failed)
shard #6 (failed)
shard #7 (failed)
shard #8 (failed)
shard #9 (failed)
missing shard #0
missing shard #1
missing shard #2
missing shard #3
missing shard #4
missing shard #5
missing shard #6
missing shard #7
missing shard #8
missing shard #9

"shards did not complete" + purple step is an indication to sheriffs etc that shard execution timed out, rather than that shards completed and failed.

I see, thanks for explaining, I'll look into them.

The telemetry_perf_unittests flakiness is tracked here:  crbug.com/836447 , both issues are being actively worked on, so I'm going to close this bug, and will file a separate bug to investigate the timeout/failure thing.