keepalived fails to start after updating DVR-HA internal network MTU
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
neutron |
In Progress
|
Medium
|
Anton Kurbatov |
Bug Description
We got an issue when keepalived stops to be running after update MTU on the internal network of the DVR-HA router.
It turned out that the keepalived config has an interface from qrouter-ns although the keepalived process itself is running in snat-ns.
Here is a simple demo on the latest master branch:
$ openstack network create net1
$ openstack subnet create sub1 --network net1 --subnet-range 192.168.100.0/24
$ openstack router create r1 --distributed --ha
$ openstack router add subnet r1 sub1
Keepalived process is running and the config looks like:
$ ps axf | grep -w pid.keepalived
...
130250 ? S 0:00 \_ keepalived -P -f /opt/stack/
$ cat /opt/stack/
global_defs {
notificatio
router_id neutron
}
vrrp_instance VR_60 {
state BACKUP
interface ha-77ee55dc-5c
virtual_
priority 50
garp_
nopreempt
advert_int 2
track_interface {
}
virtual_
}
$
Now update MTU of the internal network:
$ openstack network set net1 --mtu 1400
$ ps axf | grep -w pid.keepalived
131097 pts/0 S+ 0:00 | \_ grep --color=auto -w pid.keepalived
$
$ ip netns exec snat-f7df848f-
(/opt/stack/
(/opt/stack/
Non-existent interface specified in configuration
$
$ cat /opt/stack/
global_defs {
notificatio
router_id neutron
}
vrrp_instance VR_60 {
state BACKUP
interface ha-77ee55dc-5c
virtual_
priority 50
garp_
nopreempt
advert_int 2
track_interface {
}
virtual_
}
virtual_
}
}$
$ ip netns exec snat-f7df848f-
1: lo: <LOOPBACK,
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
10: ha-77ee55dc-5c: <BROADCAST,
link/ether fa:16:3e:46:30:c4 brd ff:ff:ff:ff:ff:ff
$
Changed in neutron: | |
importance: | Undecided → Medium |
assignee: | nobody → Anton Kurbatov (akurbatov) |
tags: | added: l3-dvr-backlog |
Related fix proposed to branch: master /review. opendev. org/c/openstack /neutron/ +/886408
Review: https:/