Improper handling of supervisord uptime output format

[ckrough]

When supervisord process uptime exceeds 24 hours, supervisor reports time in days + hh:mm:ss format. The supervisord.py script is expecting time only in hh:mm:ss format.

For example, for a running process:

mygroup:myprocess-00       RUNNING    pid 26786, uptime 6 days, 21:58:15

in _extract_uptime, it's assumed that the uptime time can be parse using

h, m, s = desc.split('uptime ')[1].split(':')

This causes the following error:

Traceback (most recent call last):
  File "/opt/datadog-agent/agent/checks/__init__.py", line 543, in run
    self.check(copy.deepcopy(instance))
  File "/etc/dd-agent/checks.d/supervisord.py", line 80, in check
    uptime = self._extract_uptime(proc)
  File "/etc/dd-agent/checks.d/supervisord.py", line 101, in _extract_uptime
    return int(s) + 60 * (int(m) + 60 * int(h))
ValueError: invalid literal for int() with base 10: '6 days, 21'

Before I update the code to handle this situation, I want to verify that this code has been used on processes running beyond 24hr and confirm that this is not a configuration issue with supervisord's output format.

@isaacdd, have you run into this?

[isaacdd]

@ckrough. I saw this and already fixed it. This is going into the dd-agent repo in the next release. For now, you can pull latest check from here: DataDog/dd-agent#1165. Please let me know if you run into any other issues.