yahoo-eng-team team mailing list archive

[Perry <panxia6679@xxxxxxxx>]

Public bug reported:

before spawning vm, neutron l2 agent has to wire up device on compute.
This is controlled by provisioning_block in neutron. However L2 agent on
compute takes much time to sync lots of fdb info during wire up device
and which causes VM takes long time to spawn and active. The fdb
synchronization could be done when spawning vm to improve the
performance of booting vm. This also delays next loop of daemon_loop to
get another new taps processed.

Steps to re-produce the problem:

L2 of linuxbridge and dedicated compute nodes

1) nova boot  --image  ubuntu-guest-image-14.04  --flavor m1.medium
--nic net-name=internal --security-groups unlocked --min-count 50 --max-
count 50 free_deletable

2)observe numerous log as below in neutron-linuxbridge-agent.log.

2017-03-07 05:01:43.220 25336 DEBUG neutron.agent.linux.utils [req-
534e9f59-0a66-4071-8c40-977f87a6be49 - - - - -] Running command:
['sudo', '/bin/neutron-rootwrap', '/etc/neutron/rootwrap.conf',
'bridge', 'fdb', 'append', '00:00:00:00:00:00', 'dev', 'vxlan-1', 'dst',
'10.160.152.141'] create_process /usr/lib/python2.7/site-
packages/neutron/agent/linux/utils.py:89

3)eventually, it takes lots of time to process new tap (6.2 seconds) in
neutron-linuxbridge-agent.log

2017-03-07 05:01:44.654 25336 DEBUG
neutron.plugins.ml2.drivers.agent._common_agent [req-835dd509-27b0-42e3
-903d-32577187d288 - - - - -] Loop iteration exceeded interval (2 vs.
6.2001709938)! daemon_loop /usr/lib/python2.7/site-
packages/neutron/plugins/ml2/drivers/agent/_common_agent.py:466

** Affects: neutron
     Importance: Undecided
         Status: New

** Description changed:

  before spawning vm, neutron l2 agent has to wire up device on compute.
- This is controlled by provision_block in neutron. However L2 agent on
+ This is controlled by provisioning_block in neutron. However L2 agent on
  compute takes much time to sync lots of fdb info during wire up device
  and which causes VM takes long time to spawn and active. The fdb
  synchronization could be done when spawning vm to improve the
  performance of booting vm. This also delays next loop of daemon_loop to
  get another new taps processed.
- 
  
  Steps to re-produce the problem:
  
  L2 of linuxbridge and dedicated compute nodes
  
  1) nova boot  --image  ubuntu-guest-image-14.04-20161130-x86_64
  --flavor m1.medium --nic net-name=internal --security-groups unlocked
  --min-count 50 --max-count 50 free_deletable
  
  2)observe numerous log as below in neutron-linuxbridge-agent.log.
  
  2017-03-07 05:01:43.220 25336 DEBUG neutron.agent.linux.utils [req-
  534e9f59-0a66-4071-8c40-977f87a6be49 - - - - -] Running command:
  ['sudo', '/bin/neutron-rootwrap', '/etc/neutron/rootwrap.conf',
  'bridge', 'fdb', 'append', '00:00:00:00:00:00', 'dev', 'vxlan-1', 'dst',
  '10.160.152.141'] create_process /usr/lib/python2.7/site-
  packages/neutron/agent/linux/utils.py:89
  
  3)eventually, it takes lots of time to process new tap (6.2 seconds) in
  neutron-linuxbridge-agent.log
  
  2017-03-07 05:01:44.654 25336 DEBUG
  neutron.plugins.ml2.drivers.agent._common_agent [req-835dd509-27b0-42e3
  -903d-32577187d288 - - - - -] Loop iteration exceeded interval (2 vs.
  6.2001709938)! daemon_loop /usr/lib/python2.7/site-
  packages/neutron/plugins/ml2/drivers/agent/_common_agent.py:466

** Description changed:

  before spawning vm, neutron l2 agent has to wire up device on compute.
  This is controlled by provisioning_block in neutron. However L2 agent on
  compute takes much time to sync lots of fdb info during wire up device
  and which causes VM takes long time to spawn and active. The fdb
  synchronization could be done when spawning vm to improve the
  performance of booting vm. This also delays next loop of daemon_loop to
  get another new taps processed.
  
  Steps to re-produce the problem:
  
  L2 of linuxbridge and dedicated compute nodes
  
- 1) nova boot  --image  ubuntu-guest-image-14.04-20161130-x86_64
- --flavor m1.medium --nic net-name=internal --security-groups unlocked
- --min-count 50 --max-count 50 free_deletable
+ 1) nova boot  --image  ubuntu-guest-image-14.04  --flavor m1.medium
+ --nic net-name=internal --security-groups unlocked --min-count 50 --max-
+ count 50 free_deletable
  
  2)observe numerous log as below in neutron-linuxbridge-agent.log.
  
  2017-03-07 05:01:43.220 25336 DEBUG neutron.agent.linux.utils [req-
  534e9f59-0a66-4071-8c40-977f87a6be49 - - - - -] Running command:
  ['sudo', '/bin/neutron-rootwrap', '/etc/neutron/rootwrap.conf',
  'bridge', 'fdb', 'append', '00:00:00:00:00:00', 'dev', 'vxlan-1', 'dst',
  '10.160.152.141'] create_process /usr/lib/python2.7/site-
  packages/neutron/agent/linux/utils.py:89
  
  3)eventually, it takes lots of time to process new tap (6.2 seconds) in
  neutron-linuxbridge-agent.log
  
  2017-03-07 05:01:44.654 25336 DEBUG
  neutron.plugins.ml2.drivers.agent._common_agent [req-835dd509-27b0-42e3
  -903d-32577187d288 - - - - -] Loop iteration exceeded interval (2 vs.
  6.2001709938)! daemon_loop /usr/lib/python2.7/site-
  packages/neutron/plugins/ml2/drivers/agent/_common_agent.py:466

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1670625

Title:
  populate fdb entries before spawning vm

Status in neutron:
  New

Bug description:
  before spawning vm, neutron l2 agent has to wire up device on compute.
  This is controlled by provisioning_block in neutron. However L2 agent
  on compute takes much time to sync lots of fdb info during wire up
  device and which causes VM takes long time to spawn and active. The
  fdb synchronization could be done when spawning vm to improve the
  performance of booting vm. This also delays next loop of daemon_loop
  to get another new taps processed.

  Steps to re-produce the problem:

  L2 of linuxbridge and dedicated compute nodes

  1) nova boot  --image  ubuntu-guest-image-14.04  --flavor m1.medium
  --nic net-name=internal --security-groups unlocked --min-count 50
  --max-count 50 free_deletable

  2)observe numerous log as below in neutron-linuxbridge-agent.log.

  2017-03-07 05:01:43.220 25336 DEBUG neutron.agent.linux.utils [req-
  534e9f59-0a66-4071-8c40-977f87a6be49 - - - - -] Running command:
  ['sudo', '/bin/neutron-rootwrap', '/etc/neutron/rootwrap.conf',
  'bridge', 'fdb', 'append', '00:00:00:00:00:00', 'dev', 'vxlan-1',
  'dst', '10.160.152.141'] create_process /usr/lib/python2.7/site-
  packages/neutron/agent/linux/utils.py:89

  3)eventually, it takes lots of time to process new tap (6.2 seconds)
  in neutron-linuxbridge-agent.log

  2017-03-07 05:01:44.654 25336 DEBUG
  neutron.plugins.ml2.drivers.agent._common_agent [req-
  835dd509-27b0-42e3-903d-32577187d288 - - - - -] Loop iteration
  exceeded interval (2 vs. 6.2001709938)! daemon_loop /usr/lib/python2.7
  /site-packages/neutron/plugins/ml2/drivers/agent/_common_agent.py:466

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1670625/+subscriptions