Skip to content

bootc-dev/bootc#2586: operator request #99

Open

Description

@jmarrero-bot

The operator commented on bootc-dev/bootc#2586, a thread outside our own space. Runs for it take this issue as their item and answer here; nothing is posted there.

jmarrero wrote (bootc-dev/bootc#2586 (comment)):

@jmarrero-bot can you review this and check the behavior on the dev space?

Generated-by: https://github.com/cgwalters/#llms

Activity

  1. jmarrero-bot commented on Nov 19, 2026

    @jmarrero-bot
    ContributorAuthor

    Triage of jmarrero's comment: claimed; started.

    Generated-by: https://github.com/cgwalters/#llms

  2. jmarrero-bot commented on Nov 19, 2026

    @jmarrero-bot
    ContributorAuthor

    ✅ dispatch ended well. Result: #99 (comment)

  3. jmarrero-bot commented on Nov 19, 2026

    @jmarrero-bot
    ContributorAuthor

    Re-checked bootc-dev/bootc#2586 at the current head d1b21bc0. The code fixes from my 2026-11-10 review (at 86a65e0e) are now in the PR, but I couldn't re-run the VM checks on the dev space: this runner has no KVM access. So the behavior on real hosts is still unconfirmed at this head.

    Each earlier failure, against the new code:

    • The /var source is now checked before install_to_filesystem touches the host (install.rs:3229-3239). The copy itself still runs after the host's boot entries are gone, and nothing checks free space first.
    • mountinfo now reads only the mountpoint field, so an empty mount source no longer breaks it (mountinfo.rs:69). The image's content at a nested mountpoint is kept (migrate.rs:1218). The destination keeps its SELinux label when the source has none (migrate.rs:1408). These are the install/migrate: Fix --preserve-var failures after the install bootc#11 fixes, so that fork PR looks superseded.
    • --preserve-var-skip now requires --preserve-var, and the config is ignored unless you preserve /var. --preserve-var also requires --experimental now.
    • On the tmt side: the plan pins ubi9/podman:9.8-1793032728, the tmpfiles rule uses printf "%s\n", and the four package-mode Testing Farm jobs pass. SSH host keys aren't fixed in code. Instead bootc-installation.7.md:372 documents the cloud-init case.

    CI on d1b21bc0 (run): three of the four failing composefs jobs are the grub-cc ones, and they fail on a koji 404 for grub2-efi-x64-cc-2.12-60.eln156, which is infra. I didn't find the cause of composefs, xfs, systemd, uki, unsealed. The tier-1 Testing Farm jobs fail at 0s. Main is green.

    Plan (here, since the PR isn't ours):

    {
      "route": "jmarrero-forge/tracker#99",
      "target": "bootc-dev/bootc#2586 (ckyrouac's branch; fixes go as a draft on top in jmarrero-forge/bootc, like jmarrero-forge/bootc#11)",
      "kind": "implement",
      "steps": [
        "Re-run the 2026-11-10 VM matrix (F0/C0/F1/C1/F2/C2/F3/C3) against d1b21bc0 on a devspace with KVM, using the documented command plus --experimental",
        "If it passes, close jmarrero-forge/bootc#11 as superseded (its six fixes are in the PR head now)",
        "Optional follow-up: preflight free space for a full (non-reflink) copy in VarMigrationPlan::prepare, since the copy still runs after the host's boot entries are removed (install.rs:3229-3231 says so)"
      ],
      "unverified": [
        "VM behavior at d1b21bc0",
        "cause of the failed systemd/uki composefs job"
      ]
    }

    Generated-by: https://github.com/jmarrero-bot#llms

    Generated by dispatch · ◷

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions