Need tuneable ovs_vsctl_timeout in ovs_agent.ini

[Drew Freiberger]

With a large amount of neutron agents we find we need to tune the ovs_vsctl_timeout in the DEFAULT section of /etc/neutron/plugins/ml2/openvswitch_agent.ini

A reasonable value on one of our clouds is 60 seconds. With a large ovs-vsctl delay, we end up with faulty states in neutron-l3 agents running active on two or three members of an ha router.

This config looks like:

[DEFAULT]
ovs_vsctl_timeout = 60

Please provide a configuration interface for this in the charm. Thank you.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/530801

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/530867

[Drew Freiberger (afreiberger)]

@gnuoy: We just realized that even hacking in this update to the /etc/neutron/plugins/ml2/openvswitch_agent.ini file, it fixes ovs agent, but does not fix neutron-l3-agent.

Can you add code into the charm-neutron-gateway project to handle appending --config-file=/etc/neutron/plugins/ml2/openvswitch_agent.ini if plugin in ['OVS', 'OVS_ODL'] into the service init file?

Note the startup command differences:

neutron 367937 8.1 0.0 569448 418396 ? Ss Jan03 104:41 /usr/bin/python /usr/bin/neutron-l3-agent --config-file=/etc/neutron/neutron.conf --config-file=/etc/neutron/l3_agent.ini --config-file=/etc/neutron/fwaas_driver.ini --log-file=/var/log/neutron/neutron-l3-agent.log

neutron 883782 22.8 0.0 511724 355836 ? Ss 2017 2312:15 /usr/bin/python /usr/bin/neutron-openvswitch-agent --config-file=/etc/neutron/neutron.conf --config-file=/etc/neutron/plugins/ml2/openvswitch_agent.ini --log-file=/var/log/neutron/neutron-openvswitch-agent.log

The neutron-l3-agent doesn't pull in the OVS configs, but DOES use ovsdb connections directly within the agent context.

[James Page (james-page)]

I'd probably just push the timeout value into the DEFAULT section of the neutron.conf file; if it applies to both n-ovs-agent and n-l3-agent thats fine.

[Drew Freiberger (afreiberger)]

popping this into neutron.conf will affect outages for conf changes in additional services I'd rather not impacted.

    NEUTRON_CONF: {
<snip>
        'services': ['neutron-l3-agent',
                     'neutron-dhcp-agent',
                     'neutron-metadata-agent',
                     'neutron-plugin-openvswitch-agent',
                     'neutron-plugin-metering-agent',
                     'neutron-metering-agent',
                     'neutron-lbaas-agent',
                     'neutron-vpn-agent']

I'd rather keep it in one conf and update the service init, or put it in the two specific conf files and limit the number of services restarted by config file refresh.

[OpenStack Infra (hudson-openstack)]

Change abandoned by Frode Nordahl (<email address hidden>) on branch: master
Review: https://review.openstack.org/530801
Reason: Marking this review as abandoned due to lack of activity in the past 6 months.
Feel free to restore it again if you want to pick up and continue the work.

[OpenStack Infra (hudson-openstack)]

Change abandoned by Frode Nordahl (<email address hidden>) on branch: master
Review: https://review.openstack.org/530867
Reason: Marking this review as abandoned due to lack of activity in the past 6 months.
Feel free to restore it again if you want to pick up and continue the work.

[Edward Hope-Morley (hopem)]

I'd like to rekindle this and update it for the more modern neutron. The config in the description of this bug refers to the older OFAgent driver which has been replaced since Newton with the "native" driver and therefore the config needed is:

[ovs]
...
of_connect_timeout = <timeout_secs>
of_request_timeout = <timeout_secs>