Close file descriptors when forking subprocess to reload haproxy

[cgbaker]

References #28

[brndnmtthws]

Makes sense, thanks!

[brndnmtthws]

Hmm, upon closer inspection, I see that close_fds=True is the default in Python 3: https://docs.python.org/3/library/subprocess.html#popen-constructor

However, close_fds=False is the default in Python 2: https://docs.python.org/2/library/subprocess.html#popen-constructor

Since marathon-lb.py is now using Python 3, I don't think we need this change. Thoughts?

[cgbaker]

Excellent point; I didn't notice that that the interpreter hint had changed to python3. Does the script still work with python2? If so, an explicit invocation, a la
python2 servicerouter.py
would still trigger the effect addressed in #28.

If it's not necessary for marathon-lb, then it seems we can close it.

I had thought (mistakenly) that the servicerouter utility was being moved out of the main marathon repo (it's not in master anymore), but I see that it's still part of the distribution in the 0.12 and 0.13 releases, where it's still coding against python2. Should I refile this over in the marathon project and get these changes integrated there? I'm not sure what the long-term plan is for this utility.

[brndnmtthws]

This is indeed the successor to servicerouter.py. It was only recently removed from the repo, however, so it probably won't be absent until the next release.

[cgbaker]

So, as the current version of servicerouter doesn't work with python 2 (because of the urllib.parse requirement) and the proposed change isn't necessary for python3, I agree that this can be closed. I'll go ahead and close it.