yahoo-eng-team team mailing list archive

[Launchpad Bug Tracker <1582602@xxxxxxxxxxxxxxxxxx>]

[Expired for neutron because there has been no activity for 60 days.]

** Changed in: neutron
       Status: Incomplete => Expired

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1582602

Title:
  part of external pid files left when ha router is deleted

Status in neutron:
  Expired

Bug description:
  Env:
  Juno 2014.2.4
  3 controller, 3 compute, 2 network(l3 agent)

  Description:
  In my Juno environment, I created a few HA routers(about 20), there would be some files generated in /var/lib/neutron/external and /var/lib/neutron/ha_confs to record keepalived/metadata proxy pid, as expected.
  But when I deleted these HA routers, part of external pid files still left.

  The external file is created by notify_master.sh, which will be executed when L3 HA state transitions to master.
  When delete the HA router, function "disable" in external_process.py will remove this file, if the pid in it is active.
  So there will be a situation:
  (1) L3 HA state transitions to master.
  (2) external pid file is created.
  (3) L3 HA state transitions to backup.
  (4) external process will be killed.
  (5) delete L3 HA
  (6) external process is not active, do not remove the pid file.
      Process for 76d8414d-3902-4511-a732-62a4759251a5 pid 21193 is stale, ignoring signal 9 disable /usr/lib/python2.7/site-packages/neutron/agent/linux/external_process.py:115
  (7) the external pid file is left.

  I am confused whether this logic(stale process ignore) is expected, or
  we should remove the external pid file whenever delete the HA router.

  By the way, the file lock for neutron-iptables in
  /var/lib/neutron/lock is left too.

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1582602/+subscriptions