All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=591704

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZymAELEhNTdG9wcGFnZUFsZXJ0UGFyZW50ImtXZWJSVEMvd2VicnRjLWxpbnV4LWxhcmdlLXRlc3RzL3dlYnJ0Y19wZXJmX3Rlc3RzL3JhbXAtdXAtZG93bi11cC1hdmVyYWdlLW5ldHdvcmstbGF0ZW5jeS9VcERvd25VcE9uZVN0cmVhbQwLEg1TdG9wcGFnZUFsZXJ0GMNbDA,agxzfmNocm9tZXBlcmZyigELEhNTdG9wcGFnZUFsZXJ0UGFyZW50Il1XZWJSVEMvd2VicnRjLWxpbnV4LWxhcmdlLXRlc3RzL3dlYnJ0Y19wZXJmX3Rlc3RzL3JhbXBfdXBfZG93bl91cF8xc3RyZWFtX25vcnR4L3NlY29uZF9yYW1wdXAMCxINU3RvcHBhZ2VBbGVydBjDWww


Bot(s) for this bug's original alert(s):

webrtc-linux-large-tests

Is this related to https://bugs.chromium.org/p/chromium/issues/detail?id=591703?

Never mind. I think I found the real smoking gun.

No, it's not  bug 591707 . This is related to switching to new machines for Linux and Mac in https://codereview.chromium.org/1662933005/, which was communicated to the sheriffs.

Quinten: this bot is clearly uploading new data successfully (https://build.chromium.org/p/client.webrtc/builders/Linux64%20Release%20%5Blarge%20tests%5D/builds/6855) but it doesn't show up here. 
I don't remember whitelisting any of these two new machines, but if I didn't - shouldn't they get an error uploading data then?

 Issue 591703  has been merged into this issue.

1. In the linked job, the that bot is uploading data for the test "webrtc_perf_tests/10_pl_10_drift", but not for webrtc_perf_tests/ramp_up_down_up_1stream_nortx/second_rampup or webrtc_perf_tests/ramp-up-down-up-average-network-latency/UpDownUpOneStream; is this expected?

2. Maybe it's possible that the two new machines have an IP address that was previously whitelisted? Searching for recent requests for data from webrtc-linux-large-tests, the requests have an IP address of 70.32.137.216, which is in dashboard IP whitelist.

Ah, thanks for figuring this out. I'm clearly jetlagged still (in KIR now, flying to MTV tonight).
Sorry for coming up with the false conclusion here, I was sure the machine switch at Feb 23 caused this, but apparently the test was disabled at the very same day.

So hlundin's dupe of  bug 591707  was correct to begin with.