fix(backups): disable CBT when CBT is unusable

[fbeauchamp]

Description

Short explanation of this PR (feel free to re-use commit message)

Checklist

• Commit
  • Title follows commit conventions
  • Reference the relevant issue (Fixes #007, See xoa-support#42, See https://...)
  • If bug fix, add Introduced by
• Changelog
  • If visible by XOA users, add changelog entry
  • Update "Packages to release" in CHANGELOG.unreleased.md
• PR
  • If UI changes, add screenshots
  • If not finished or not tested, open as Draft

Review process

This 2-passes review process aims to:

• develop skills of junior reviewers
• limit the workload for senior reviewers
• limit the number of unnecessary changes by the author

1. The author creates a PR.
2. Review process:
   1. The author assigns the junior reviewer.
   2. The junior reviewer conducts their review:
      • Resolves their comments if they are addressed.
      • Adds comments if necessary or approves the PR.
   3. The junior reviewer assigns the senior reviewer.
   4. The senior reviewer conducts their review:
      • If there are no unresolved comments on the PR → merge.
      • Otherwise, we continue with 3.
3. The author responds to comments and/or makes corrections, and we go back to 2.

Notes:

1. The author can request a review at any time, even if the PR is still a Draft.
2. In theory, there should not be more than one reviewer at a time.
3. The author should not make any changes:
   • When a reviewer is assigned.
   • Between the junior and senior reviews.

[rtjdamen]

Hi @fbeauchamp, i want to comment on this one. It’s not needed to disable cbt on the vdi itself, just removing all cbt snapshots and run a new full is enough to resolve the problems with invalid cbt here. We have seen some rare issues with disabling CBT on a running VM causing the vdi to get stucked into paused state. So i think it’s better to not run that command unsupervised without knowing what can happen.

[fbeauchamp]

Hi @fbeauchamp, i want to comment on this one. It’s not needed to disable cbt on the vdi itself, just removing all cbt snapshots and run a new full is enough to resolve the problems with invalid cbt here. We have seen some rare issues with disabling CBT on a running VM causing the vdi to get stucked into paused state. So i think it’s better to not run that command unsupervised without knowing what can happen.

this won't be run automatically, at least for now. It will be used when manually disabling Changed Block Tracking , as an additional way to clean up any remaining data, broken cbt chain.
Also calling disable_cbt on the xapi is the only way to make it remove the *.cbtlog file without accessing the storage repository

[b-Nollet]

Looks good to me apart from the little typos in the comments

[rtjdamen]

After installing this update we ran into this error on a cvm "SR_BACKEND_FAILURE_82(, Failed to snapshot VDI [opterr=['MAP_DUPLICATE_KEY', 'VDI', 'sm_config', 'OpaqueRef:69fc25e1-45a5-4aa1-80af-365773dd2fa9', 'paused']], )" i have seen it before when u try to disable cbt on running vms. this happened from the backup job. Seems like this has been initiated because the cbt for this vm was invalid.