multiple engines, multiple containers, same pid

[zzzeek]

we see this in nova, multiple sets of values:

[2019-11-18 23:46:52] [sqlalchemy_collectd.server.plugin] receive[UDP:localhost:25827] -> sqlalchemy_collectd.Values(type='sqlalchemy_totals', type_instance='6', plugin='sqlalchemy', plugin_instance='nova', host='overcloud-controller-0', time=1574120812, interval=2, values=[328630, 0, 457, 456])
[2019-11-18 23:46:53] [sqlalchemy_collectd.server.plugin] receive[UDP:localhost:25827] -> sqlalchemy_collectd.Values(type='sqlalchemy_totals', type_instance='6', plugin='sqlalchemy', plugin_instance='nova', host='overcloud-controller-0', time=1574120813, interval=2, values=[164346, 0, 457, 455])
[2019-11-18 23:46:53] [sqlalchemy_collectd.server.plugin] receive[UDP:localhost:25827] -> sqlalchemy_collectd.Values(type='sqlalchemy_totals', type_instance='6', plugin='sqlalchemy', plugin_instance='nova', host='overcloud-controller-0', time=1574120813, interval=2, values=[164346, 0, 457, 455])
[2019-11-18 23:46:54] [sqlalchemy_collectd.server.plugin] receive[UDP:localhost:25827] -> sqlalchemy_collectd.Values(type='sqlalchemy_totals', type_instance='6', plugin='sqlalchemy', plugin_instance='nova', host='overcloud-controller-0', time=1574120813, interval=2, values=[283516, 0, 907, 905])
[2019-11-18 23:46:54] [sqlalchemy_collectd.server.plugin] receive[UDP:localhost:25827] -> sqlalchemy_collectd.Values(type='sqlalchemy_totals', type_instance='6', plugin='sqlalchemy', plugin_instance='nova', host='overcloud-controller-0', time=1574120814, interval=2, values=[328630, 0, 457, 456])
[2019-11-18 23:46:55] [sqlalchemy_collectd.server.plugin] receive[UDP:localhost:25827] -> sqlalchemy_collectd.Values(type='sqlalchemy_totals', type_instance='6', plugin='sqlalchemy', plugin_instance='nova', host='overcloud-controller-0', time=1574120815, interval=2, values=[164346, 0, 457, 455])
[2019-11-18 23:46:55] [sqlalchemy_collectd.server.plugin] receive[UDP:localhost:25827] -> sqlalchemy_collectd.Values(type='sqlalchemy_totals', type_instance='6', plugin='sqlalchemy', plugin_instance='nova', host='overcloud-controller-0', time=1574120815, interval=2, values=[164346, 0, 457, 455])
[2019-11-18 23:46:56] [sqlalchemy_collectd.server.plugin] receive[UDP:localhost:25827] -> sqlalchemy_collectd.Values(type='sqlalchemy_totals', type_instance='6', plugin='sqlalchemy', plugin_instance='nova', host='overcloud-controller-0', time=1574120815, interval=2, values=[283516, 0, 907, 905])

two different sets of totals and same pid of 6. now i would think this is because, OK nova has the main DB and the API DB, but looking in plugin.py, CollectionTarget is a process-wide singleton, multiple engines go into it. i see there are two Senders set up, which is wrong and we can see messages being sent twice above, but...they send on the same collection target which should mean only one value. the above are running in containers so "6" is local to a container, need to see if that is a factor here.

[zzzeek]

so if you look in the podman containers for all 8 nova services, they are all on pid 6 :) which means we have an issue here that these are different processes but they see the same pid due to containers :)

[zzzeek]

and im going to take a guess that running in a docker container, you have a pid, the job of "hey what's my pid on the host machine?" is exactly where "sandbox" would say, GET LOST YOU'RE IN A SANDBOX WE'RE TRYING TO PRETEND THIS IS SECURE! . sooooo.....pid + hostname is not enough! we need something else for containers with net=host. bleah.