Comment 9 for bug 1654287

Reviewed: https://review.openstack.org/421325
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=3f9f740d81b51be5e563069c720366fa90ade9ee
Submitter: Jenkins
Branch: master

commit 3f9f740d81b51be5e563069c720366fa90ade9ee
Author: Daniel Alvarez <email address hidden>
Date: Thu Jan 12 01:06:01 2017 +0000

    Fix netns_cleanup interrupted on rwd I/O

    Functional tests for netns_cleanup have been failing a few times
    in the gate lately. After thorough tests we've seen that the issue was
    related to using rootwrap-daemon inside a wait_until_true loop. When
    timeout fired while utils.execute() was reading from rootwrap-daemon,
    it got interrupted and the output of the last command was not read.
    Therefore, next calls to utils.execute() would read the output of
    their previous command rather than their own, leading to unexpected
    results.

    This fix will poll existing processes in the namespace without making
    use of the wait_until_true loop. Instead, it will check elapsed time
    and raise the exception if timeout is exceeded.

    Also, i'm removing debug traces introduced in
    327f7fc4d54bbaaed3778b5eb3c51a037a9a178f which helped finding the root
    cause of this bug.

    Change-Id: Ie233261e4be36eecaf6ec6d0532f0f5e2e996cd2
    Closes-Bug: #1654287