Don't report android device port paths.

infra/services/sysmon/android_device_metrics.py

Index: infra/services/sysmon/android_device_metrics.py
diff --git a/infra/services/sysmon/android_device_metrics.py b/infra/services/sysmon/android_device_metrics.py
index 8616be068bb5d0a5f1a777074ecd6c4ff5d2851a..6990cc0c19b80fcd22e3b585f3dbf9c28a86b6c7 100644
--- a/infra/services/sysmon/android_device_metrics.py
+++ b/infra/services/sysmon/android_device_metrics.py
@@ -5,6 +5,7 @@
 import json
 import logging
 import os
+import re
 import socket
 import time
 
@@ -23,6 +24,8 @@ ANDROID_DEVICE_FILE = os.path.join(os.path.expanduser('~'),
 # Don't read a file older than this many seconds.
 ANDROID_DEVICE_FILE_STALENESS_S = 120
 
+PORT_PATH_RE = re.compile(r'\d+\/\d+')
+
 
 cpu_temp = ts_mon.FloatMetric('dev/cpu/temperature',
                               description='device CPU temperature in deg C')
@@ -51,6 +54,11 @@ def get_device_statuses(device_file=ANDROID_DEVICE_FILE, now=None):
     return
 
   for device_name, device in devices.iteritems():
+    if PORT_PATH_RE.match(device_name):
+      logging.warning('Found port path %s as device id. Skipping.',

[ghost stip (do not use), 2016/06/01 18:03:37]

interesting. when would this come up instead of device? (I guess that might be a
swarming thing?) if we drop it, are we losing data?

[bpastene, 2016/06/01 21:43:30]

I imagine it happens when there's a protocol/usb failure with the device. It
happens a bunch to this guy (look at all the USBErrorIO errors):
https://chromium-swarm.appspot.com/restricted/bot/build342-m4
When it does happen, the only data we have is 'status: unavailable', which we're
unable to even map to a specific device id. So no, we are not.

+                      device_name)
+      continue
+
     fields = {'device_id': device_name}
 
     # Fields with special handling.