Seems just failing on Android Nexus5X Perf (3) now. And the failure pattern seems to be more like flaky than failure to me. 

Can any of the owner take a look?

+cc: jbroman@

For the first link, structured clone's tests are failing very often.  Any ideas?

Hmm, timing out. I'm not sure what's going on, though it does seem as though the new 5X (see  bug 699719 ) has less stable performance than its predecessor:

https://chromeperf.appspot.com/report?sid=a6d05b7150089c4f714c84be33c8734954cc724c46d346266ea95dd2de19b6e2

It previously gave results that were reliably 27-28, but now seems to spike to 63. While this is still much faster than this test used to run, it makes me wonder whether something's up with this device. Aside from that, is it possible to get the console output from Telemetry tests on bots, so that we can at least double-check that this isn't timing out due to an exception being thrown somewhere?

It looks like the last 4 builds were green!

I just double-checked and there is just the single device swap on the perf dashboard at 699719, so although it does look like the new device is slower/noisier there haven't been more device swaps.

It also looks like the spikes on the dashboard don't fully correspond to failures (sometimes the value is ~60 and it's passing, sometimes failing) so I'm not sure if the timeout failure is related to the test running slower.

For console output, you mean the JS console output? I think it is blank in this log: https://luci-logdog.appspot.com/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_Nexus5X_Perf__3_%2F3560%2F%2B%2Frecipes%2Fsteps%2Fblink_perf.bindings%2F0%2Fstdout

+Ned in case he knows whether this works properly on Android

The JS console output should be there in case the failure is related to JS execution.

Ah, I clearly didn't search for the right term in the log. Yeah, no console output, so I'm not sure what's up (if serialization encountered a problem, I'd expect a RangeError or DataCloneError to be logged).

Friendly perfbot sheriff ping.

This is no longer happenning.