Restarting systemd-networkd breaks keepalived clusters

Bug #1820281 reported by sascha arthur on 2019-03-15
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
keepalived (Ubuntu)
Undecided
Unassigned
netplan.io (Ubuntu)
High
Unassigned
systemd (Ubuntu)
High
Unassigned

Bug Description

Hello,

Managing a HA floating IP with netplan and keepalived brings trouble.
Im managing a two node keepalived cluster with a floating IP.
Everything works out except if i run on a the master node (which holds floating ip) "netplan apply".
Prob netplan resets the whole interface, leaving keepalived in a broken state.
This works without any trouble using the old iproute2 approach.

Even setting, net.ipv4.conf.*.promote_secondaries does not help out.

We need definitly an non strict reset option for interfaces in thise case.

sascha arthur (blonkel) on 2019-03-15
summary: - netplan, keepalived
+ netplan, keepalived, netplan-apply
summary: - netplan, keepalived, netplan-apply
+ Restarting systemd-networkd breaks keepalived clusters
Changed in systemd (Ubuntu):
importance: Undecided → High
Changed in netplan.io (Ubuntu):
importance: Undecided → High
status: New → Confirmed
Changed in systemd (Ubuntu):
status: New → Confirmed
sascha arthur (blonkel) wrote :

Heres a possible workaround until its solved (did not test it, but seems valid): https://chr4.org/blog/2019/01/21/make-keepalived-play-nicely-with-netplan-slash-systemd-network/

Personally i can live with 1-2 sek downtime (on network change), thats why i added a subscribe hook in puppet to restart keepalived.

Will get rid of the workaround as soon as its solved, which is currently hardly worked on it seems.

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers