yahoo-eng-team team mailing list archive

[Lajos Katona <1618244@xxxxxxxxxxxxxxxxxx>]

If I understand well the above patch fixed the issue, please don't
hesitate to reopen this bug if you see the scale issue

** Changed in: neutron
       Status: In Progress => 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/1618244

Title:
  Possible scale issues with neutron-fwaas requesting all tenants with
  firewalls after RPC failures

Status in neutron:
  Fix Released

Bug description:
  Information zzelle in conversation with njohnston

  An overload is caused first by some neutron-servers crashed, secondly
  by every l3-agent trying to perform a "full" process_services_sync.
  When we restarted every crashed neutron-servers and purge neutron
  queues, we restarted crashed neutron-servers rpc workers are still
  overloaded because of full syncs.

  About 60 L3Agents, with one router per L3Agent.

  Key question: typically i don't understand why in full sync a
  l3-agents request tenants with FWs intead of requesting its tenants
  with FW ?

  https://github.com/openstack/neutron-
  fwaas/blob/master/neutron_fwaas/services/firewall/agents/l3reference/firewall_l3_agent.py#L224

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1618244/+subscriptions