Consider a separate charm-contrail-service-checks
Bug #1881127 reported by
Adam Dyess
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
charm-openstack-service-checks |
Won't Fix
|
Wishlist
|
Unassigned |
Bug Description
Contrail exposes many API endpoints from various ports. Some can be http or https, but the service checks surrounding those are limited.
perhaps a separate charm to check specifically for contrail apis is worth while.
here's a few api endpoints to support:
keepalived
bind-definitions
contrail ui dashboard
To post a comment you must log in.
The current deployment has: 8082/tcp, 8143/tcp Unit is ready policy- routing/ 9 active idle 10.X.X.X Unit ready client/ 54 active idle 10.X.X.X System successfully registered
"""
contrail-haproxy/0* active idle 6/lxd/0 10.X.X.X 8081/tcp,
external-
keepalived/0 active idle 10.X.X.X VIP ready
landscape-
nrpe-container/12 active idle 10.X.X.X icmp,5666/tcp ready
"""
* keepalived has port 8143, and is related to contrail-haproxy (contrail- haproxy: juju-info <-> keepalived: juju-info)
* contrail-haproxy is related to contrail-controller over http-services (8082) and https-services (8143).
* contrail-haproxy is related to contrail-analytics over http-services (8081)
OTOH, -service- checks can already monitor port 8081 if contrail_ analytics_ vip Juju parameter is set to the keepalive VIP.
* charm-openstack
We would need something similar to the OpenStack API endpoint nrpe checks, via the check_http tool. Checking if "-e Unauthorized" is found should be enough.
As suggested by Adam, having charm-contrail- service- checks would avoid polluting charm-openstack -service- checks.