yahoo-eng-team team mailing list archive

[Eugene Nikanorov <enikanorov@xxxxxxxxxxxx>]

Public bug reported:

For some reason simple GET access to a floating IP on a 200-node cluster
during boot_runcommand_delete test takes too much time, ranging from 0
to 35 seconds with median ~15 seconds.

That leads to timeouts on nova side.
Also, some traces in RPC workers appear: http://paste.openstack.org/show/474339/
That may indicate that the table is a contention point.
Slowdown may be caused by constant retries.

** Affects: neutron
     Importance: High
     Assignee: Eugene Nikanorov (enikanorov)
         Status: Confirmed


** Tags: scale

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1500430

Title:
  Access to floating IP can take 30+ seconds during
  boot_runcommand_delete test on scale.

Status in neutron:
  Confirmed

Bug description:
  For some reason simple GET access to a floating IP on a 200-node
  cluster during boot_runcommand_delete test takes too much time,
  ranging from 0 to 35 seconds with median ~15 seconds.

  That leads to timeouts on nova side.
  Also, some traces in RPC workers appear: http://paste.openstack.org/show/474339/
  That may indicate that the table is a contention point.
  Slowdown may be caused by constant retries.

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1500430/+subscriptions