Add: Prevent skipping local storage

[wak327]

Fixes #862

#1034 covered the networking half of this issue. This PR covers the storage half, using the same approach.

If the existing cluster already has the local storage pool, the new members must also be set up with a disk for it. Right now microcloud add still asks Would you like to set up local storage?, and answering no causes errors later in the setup. With this change:

• If the cluster already has the local pool and we are not bootstrapping, the yes/no question is skipped and the disk selection table is shown straight away. The table already requires exactly one disk per new system.
• If any new system has no available disk, microcloud add now fails right away with Cannot set up the existing local storage pool on all new systems. Some systems do not have an available disk. Before, it printed a warning and skipped the pool.

Nothing changes for microcloud init. Nothing changes for microcloud service add either, since it runs with bootstrap: true. Adding to a cluster that doesn't have the local pool still asks the question as before.

Test suite: following the SETUP_OVN_EXPLICIT/SETUP_OVN_IMPLICIT split from #1034, this adds SETUP_ZFS_IMPLICIT, which skips the yes/no answer but still selects the disks. The add sessions that join a cluster which already has the local pool now use it: two in test/suites/add.sh and two in test/suites/basic.sh. I checked that microcloud_interactive produces the same input as before for SETUP_ZFS=yes, and the same input minus the yes line for SETUP_ZFS_IMPLICIT=yes. I couldn't run the LXD-based system suites locally, so CI will be the first full run.

Checklist

• I have read the contributing guidelines and attest that all commits in this PR are signed off, cryptographically signed, and follow this project's commit structure.
• I have checked and added or updated relevant documentation.

[wak327]

@roosterfish this follows #1034 for the storage half of #862: when the cluster already has the local pool, add no longer asks whether to set up local storage, and it fails early if a new member has no available disk. The system tests get a matching SETUP_ZFS_IMPLICIT. Could you take a look?

[Copilot]

Copilot review overview

🔵 Needs a closer look

Storage setup behavior warrants final human review.

Review effort: Lite
Findings: None

What changed in this PR

Updates microcloud add to require local-storage disks when joining a cluster with an existing local pool.

Changes:

• Enforces disk availability for all new members.
• Skips the redundant storage prompt during applicable joins.
• Updates test helpers and integration scenarios.

File	Description
test/​suites/​basic.sh	Updates cluster-growth scenarios.
test/​suites/​add.sh	Updates add-session coverage.
test/​includes/​microcloud.sh	Supports implicit ZFS disk selection.
cmd/​microcloud/​ask.go	Enforces mandatory local disk selection for joins.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[roosterfish]

I was giving it a try and this already looks good. But after removing the question whether or not to configure local storage, it should be clear for the user what the table is about when getting prompted to select a disk:

Gathering system information...
Select exactly one disk from each cluster member:          
 Filter | <type to filter rows>
   ┌──────────┬───────────────┬──────────┬──────┬────────────────────────────────────────────────────┐
   │ LOCATION │     MODEL     │ CAPACITY │ TYPE │                        PATH                        │
   ├──────────┼───────────────┼──────────┼──────┼────────────────────────────────────────────────────┤
 > │ micro02  │ QEMU HARDDISK │ 10.00GiB │ scsi │ /dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_disk1 │
   │ micro02  │ QEMU HARDDISK │ 10.00GiB │ scsi │ /dev/disk/by-id/scsi-0QEMU_QEMU_HARDDISK_lxd_disk2 │
   └──────────┴───────────────┴──────────┴──────┴────────────────────────────────────────────────────┘
 space to select; enter to confirm
 ↑/↓ to move; → to select all; ← to select none

So what about changing the sentence Select exactly one disk from each cluster member: to Select exactly one disk from each cluster member for local storage:?

[wak327]

@roosterfish thanks, good point. The prompt now reads Select exactly one disk from each cluster member for local storage: (954fee6), and the multi-member tutorial output is updated to match (c06b1a3). Could you take another look?

[roosterfish]

Thanks, looks good to me. I'll merge once we are done cutting the branch for the upcoming release cycle.

[roosterfish]

You can rebase with main to unblock the docs test.

[wak327]

@roosterfish I rebased onto main to unblock the docs check and addressed both inline comments: the error now names the affected system using the MicroOVN wording, and there's a new system test that triggers it. Could you take another look?

[roosterfish]

Please also squash the two new commits with the existing ones that add the new error message/tests so that you don't add and remove changes in the same PR.

[wak327]

@roosterfish I switched the test to lxc storage volume detach as you suggested and squashed the follow-ups into the original commits: the error message change is now part of 564100e and the new test is part of dc00e49, so nothing gets added and then changed within the PR. Could you take another look?

[roosterfish]

One more comment after reviewing the added test.

[wak327]

@roosterfish I've removed the unnecessary snap toggling for micro04 in the new test and squashed it into the existing test commit. Could you take another look?

[roosterfish]

Looks good to me!