OperationalError on charm deployment in k8s model

[Tom Haddon]

I'm using:

microk8s v1.19.3 1774 latest/edge canonical✓ classic

juju 2.9-rc2

I just tried the following in a new model:

juju deploy cs:~postgresql-charmers/postgresql-k8s postgresql
juju deploy cs:~mattermost-charmers/mattermost --config juju-external-hostname=foo.internal
juju add-relation mattermost postgresql:db
juju expose mattermost

I got the following https://paste.ubuntu.com/p/3W7dTQxXFj/

Note that I ran the deploy and relate commands pretty soon after each other, didn't wait for the charms to complete deployment.

[Tom Haddon (mthaddon)]

I tried again, this time waiting until the initial deployment of each charm had completed, but still got the same error. Checking the postgresql-0 pod directly I saw https://paste.ubuntu.com/p/jy3P8xv76x/

[Stuart Bishop (stub)]

I'll need to see the pod log or /var/log/postgresql/repmgr.log to see what failed. https://paste.ubuntu.com/p/jy3P8xv76x/ shows repmgr setup has hung. Possibly dns isn't working on this microk8s (perhaps the feature not enabled?)

[Tom Haddon (mthaddon)]

For the record, https://paste.ubuntu.com/p/Hxmk3hd2nk/.

Having chatted outside of this bug, sounds like there are some Juju 2.9 issues that you're working through currently.

[Stuart Bishop (stub)]

https://discourse.charmhub.io/t/postgresql-k8s-charm/3765/3

[Tom Haddon (mthaddon)]

Yup, confirmed it works for me now.