[2023/stable] cinder integration to microceph not firing on single node

[Matt Verran]

When installing on single node i've always installed with --accept-defaults to start with to get the minimal install up and running. This works fine now about 99.999% of the time.

Node has been bootstrapped with roles: control, compute

What I then do is rerun bootstrap to add the storage role where I manually accept in the various variables (i know i should be using preseed here).

Management networks shared by hosts (CIDRs, separated by comma) (192.168.68.0/24):
MetalLB address allocation range (supports multiple ranges, comma separated) (10.20.21.10-10.20.21.20):
Disks to attach to MicroCeph (/dev/disk/by-id/wwn-0x50026b7381077d10,/dev/disk/by-id/wwn-0x50026b7381077cde,/dev/disk/by-id/wwn-0x50026b7381077d0b,/dev/disk/by-id/ata-hp_TSST_DVDROM_SU-108HB_S1BL6YDH1004SK): /dev/disk/by-id/wwn-0x50026b7381077d10,/dev/disk/by-id/wwn-0x50026b7381077cde,/dev/disk/by-id/wwn-0x50026b7381077d0b
Node has been bootstrapped with roles: storage, compute, control
update.go:85: cannot change mount namespace according to change mount (/run/user/1000/doc/by-app/snap.juju /run/user/1000/doc none bind,rw,x-snapd.ignore-missing 0 0): cannot inspect "/run/user/1000/doc": lstat /run/user/1000/doc: permission denied
sunbeam-controller:mz640/openstack (no change)

The ceph offer is there but hasn't been integrated - I will reinstall and see if its purely due to the double bootstrap.

Given the mitigation is to simply run juju integrate admin/controller.microceph cinder-ceph this isn't a major problem.

snap at time of report [2023.2/stable (335)]

[Matt Verran (mv-2112)]

Confirmed. Double 'bootstrapping' on a single node install will cause the microceph relation to not trigger when storage role is added.

When run once only with 'sunbeam cluster bootstrap --role storage --role compute' microceph integrates as it should.

[Matt Verran (mv-2112)]

Suggest close as user error unless you see an enhancement to behaviour.