neutron dhcp agent looks for external pid unexpectedly

[Brent Eagles]

From the debug logs, it looks like the dhcp agent continually checks for an pid file for an external process that is not expected. e.g.:

DEBUG neutron.agent.linux.utils [-] Unable to access /opt/stack/data/neutron/external/pids/17951cd9-d28e-4ea9-9238-f97652690c59.pid {{(pid=10803) get_value_from_file /opt/stack/neutron/n>

This might be expected if force_metadata were true, but as per default config it is false in these deployments.

Is this expected behavior (e.g. part of some auto-cleanup from config change logic, etc.)? If so it might be useful to add some logging to that effect.

[pandatt (pandatt)]

Hi Brent, had your devstack-machine(or neutron-dhcp-agent service) been restarted when the debug messages occured ?

[Slawek Kaplonski (slaweq)]

After some analysis I don't think it is a bug. When new dnsmasq process is going to be spawned, it calls pm.enable() method here: https://github.com/openstack/neutron/blob/master/neutron/agent/linux/dhcp.py#L469

It then ends in https://github.com/openstack/neutron/blob/master/neutron/agent/linux/external_process.py#L142 when it tries to check process' pid to check if process is active. As it is new network and process is not active yet, it's normal that such debug message is logged.