deployed instances lose dns after performing ovs to ovn upgrade
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Neutron Open vSwitch Charm |
New
|
Undecided
|
Unassigned |
Bug Description
While performing an OVS to OVN upgrade, all my instances (Focal) which were deployed on top of the Openstack got a weird DNS entry after the upgrade was finished.
The deployment was running with designate/
from an instance running
$ resolvectl dns
Global:
Link 2 (ens3): 10.246.167.120 10.246.166.57
after the migration:
$ resolvectl dns
Global:
Link 2 (ens3): 172.16.0.10 172.16.0.11
but if I tell the instance to do something using a designate-bind address its OK:
$ dig ntp.ubuntu.com @10.246.167.120
; <<>> DiG 9.16.1-Ubuntu <<>> ntp.ubuntu.com @10.246.167.120
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 47523
;; flags: qr rd ra; QUERY: 1, ANSWER: 4, AUTHORITY: 0, ADDITIONAL: 1
;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 4096
; COOKIE: 5726a28294aafe6
;; QUESTION SECTION:
;ntp.ubuntu.com. IN A
;; ANSWER SECTION:
ntp.ubuntu.com. 390 IN A 91.189.94.4
ntp.ubuntu.com. 390 IN A 91.189.89.198
ntp.ubuntu.com. 390 IN A 91.189.89.199
ntp.ubuntu.com. 390 IN A 91.189.91.157
After some time (over night) the instances recovered with the proper DNS resolution as before the migration. I expect this is the result of the instance getting a new DHCP lease with the right information.
kuberentes (instances ontop of Openstack) while having a weird dns crashdump