Bootstrapping with disks in preseeds while actually missing fails with cryptic error

[Guillaume Boutry]

I bootstrapped a node running: `sunbeam cluster bootstrap -p preseed.yaml --role storage --role compute --role control` on a VM without disks attached.

With disks defined in preseed.yaml as:

microceph_config:
  bm0.lxd:
    osd_devices: /dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_osd0,/dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_osd1,/dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_osd2

Running without verbose ends with:

Error: {'return-code': 0}

Running with verbose, you get these logs:

           DEBUG Running action list-disks on microceph/0 microceph.py:269
[08:28:02] DEBUG Result after running action list-disks: {'osds': '[]', 'return-code': 0, 'stdout': "{'osds': [], 'unpartitioned-disks': []}\n", 'unpartitioned-disks': '[]'} microceph.py:273
           DEBUG Unpartitioned disks: [] microceph.py:286
           DEBUG {'microceph_config': {'bm0.lxd': {'osd_devices': '/dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_osd0,/dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_osd1,/dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_osd2'}}} microceph.py:319
           DEBUG [put] http+unix://%2Fvar%2Fsnap%2Fopenstack%2Fcommon%2Fstate%2Fcontrol.socket/1.0/config/TerraformVarsMicroceph, args={'data': '{"microceph_config": {"bm0.lxd": {"osd_devices": service.py:118
                    "/dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_osd0,/dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_osd1,/dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_osd2"}}}'}
           DEBUG http://localhost:None "PUT /1.0/config/TerraformVarsMicroceph HTTP/1.1" 200 108 connectionpool.py:456
           DEBUG Response(<Response [200]>) = {"type":"sync","status":"Success","status_code":200,"operation":"","error_code":0,"error":"","metadata":{}} service.py:120

           DEBUG Running step bm0.lxd common.py:241
           DEBUG Running action add-osd on microceph/0 microceph.py:350
⠸ Configuring MicroCeph storage ... [08:28:03] DEBUG {'return-code': 0} microceph.py:363
           DEBUG Finished running step 'bm0.lxd'. Result: ResultType.FAILED common.py:244
Error: {'return-code': 0}

After attaching the disks to the VM, the bootstrap ran to completion. Error message could be better on this error case.

[Hemanth Nakkina (hemanth-n)]

PR proposed: https://github.com/canonical/snap-openstack/pull/190

[Hemanth Nakkina (hemanth-n)]

Actions in charm-microceph returns error message outside of results dict

$ juju run microceph/0 add-osd device-id=/dev/sdb --format json
Running operation 27 with 1 task
  - task 28 on unit-microceph-0

Waiting for task 28...
{"microceph/0":{"id":"28","message":"Error: failed to record disk: This \"disks\" entry already exists\n","results":{"return-code":0},"status":"failed","timing":{"completed":"2024-05-23 04:01:21 +0000 UTC","enqueued":"2024-05-23 04:01:20 +0000 UTC","started":"2024-05-23 04:01:20 +0000 UTC"},"unit":"microceph/0"}}

It will be useful if the message is part of results dict since libjuju get_action_output returns just the results section.

[Hemanth Nakkina (hemanth-n)]

Proposed PR: https://github.com/canonical/charm-microceph/pull/75