yahoo-eng-team team mailing list archive

[Ryan Beisner <1414218@xxxxxxxxxxxxxxxxxx>]

This bug was fixed in the package neutron - 1:2014.1.5-0ubuntu5~cloud0
---------------

 neutron (1:2014.1.5-0ubuntu5~cloud0) precise-icehouse; urgency=medium
 .
   * New update for the Ubuntu Cloud Archive.
 .
 neutron (1:2014.1.5-0ubuntu5) trusty; urgency=medium
 .
   * Backport performance fix by refactoring logging statements. (LP: #1414218):
     - d/p/refactor-log-in-loop.patch: do not perform debug trace with each
       iteration through the loop of ports, instead log it once at the end.


** Changed in: cloud-archive/icehouse
       Status: Fix Committed => Fix Released

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1414218

Title:
  Remove extraneous trace in linux/dhcp.py

Status in Ubuntu Cloud Archive:
  Invalid
Status in Ubuntu Cloud Archive icehouse series:
  Fix Released
Status in neutron:
  Fix Released
Status in neutron juno series:
  Fix Released
Status in neutron package in Ubuntu:
  Fix Released
Status in neutron source package in Trusty:
  Fix Released

Bug description:
  [Impact]

  The debug tracepoint in Dnsmasq._output_hosts_file is extraneous and
  causes unnecessary performance overhead when creating lots (> 1000)
  ports at one time.

  The trace point is unnecessary since the data is being written to disk
  and the file can be examined in a worst case scenario. The added
  performance overhead is an order of magnitude in difference (~.5
  seconds versus ~.05 seconds at 1500 ports).

  [Test Case]

  1. Deploy OpenStack using neutron for networking
  2. Create 1500 ports
  3. Observe the performance degradation for each port creation.

  [Regression Potential]

  Minimal. This code has been running in stable/juno, stable/kilo, and
  above for awhile.

  [Other Questions]

  This is likely to occur in OpenStack deployments which have large
  networks deployed. The degradation is gradual, but the performance
  becomes unacceptable with large enough networks.

To manage notifications about this bug go to:
https://bugs.launchpad.net/cloud-archive/+bug/1414218/+subscriptions