Add new methods to manage instance server (reinstall, backup create, info, download, delete

[aliel]

related to:
aleph-im/aleph-sdk-python#279
aleph-im/aleph-vm#874

[github-actions]

Failed to retrieve llama text: POST 500: {"error": "500: Unhandled error during initialisation"}

[codecov]

Codecov Report

❌ Patch coverage is 19.16667% with 194 lines in your changes missing coverage. Please review.
✅ Project coverage is 58.29%. Comparing base (58a1afc) to head (e88dc25).

Files with missing lines	Patch %	Lines
src/aleph_client/commands/instance/backup.py	15.38%	176 Missing ⚠️
src/aleph_client/commands/instance/__init__.py	43.75%	18 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #437      +/-   ##
==========================================
- Coverage   60.70%   58.29%   -2.42%     
==========================================
  Files          21       22       +1     
  Lines        4057     4287     +230     
  Branches      606      642      +36     
==========================================
+ Hits         2463     2499      +36     
- Misses       1323     1517     +194     
  Partials      271      271              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[odesenfans]

Don't you risk looping forever? Why not call the corresponding GET method here instead?

[odesenfans]

Also, no timeout.

[odesenfans]

Missing metavar=metavar_valid_chains on all appearances of chain in this file.

[odesenfans]

No user confirmation required for this one?

[foxpatch-aleph]

The PR adds useful backup/restore/reinstall commands with good UX (progress bars, warnings, confirmation prompts). However, there is one functional bug that makes the --background polling loop create new backups on every poll iteration instead of checking status, and a subtle error-handling bug in reinstall where return 1 silently succeeds from Typer's perspective. There are also no tests for any of the new functionality.

src/aleph_client/commands/instance/backup.py (line 101): Bug: polling loop re-invokes create_backup instead of checking status. Each iteration of the while status == 202 loop calls manager.create_backup(...) again, which will attempt to start a new backup rather than poll the status of the one already running. This should call a status/get endpoint (e.g. manager.get_backup(vm_id=vm_id)) to check whether the in-progress backup has completed.

# Should be something like:
status, result = await manager.get_backup(vm_id=vm_id)

src/aleph_client/commands/instance/__init__.py (line 1170): Bug: return 1 does not set the process exit code in a Typer command. Typer ignores integer return values from command functions; return 1 here exits the function normally with exit code 0. Every other error path in this PR uses raise typer.Exit(1) — use that here for consistency and correctness:

raise typer.Exit(1)

src/aleph_client/commands/instance/backup.py (line 209): Partial output file left on disk after a failed/interrupted download. If the connection drops mid-transfer, output will contain a partial file with no indication to the user. Consider deleting the output file in an error handler, or writing to a temp file and renaming on success:

tmp = output.with_suffix(output.suffix + '.tmp')
try:
    with open(tmp, 'wb') as f:
        ...
    tmp.rename(output)
except:
    tmp.unlink(missing_ok=True)
    raise

src/aleph_client/commands/instance/backup.py (line 293): Missing confirmation prompt before destructive delete operation. All other destructive commands (reinstall, restore) show a warning and ask for confirmation. delete silently deletes the backup — a safety net the user may need — without any prompt. Consider adding a yes_no_input("Delete backup {backup_id}?", default=False) guard.

src/aleph_client/commands/instance/backup.py (line 404): ProgressFile class defined inside a with block inside a function. This is hard to read, impossible to unit-test independently, and triggers linter warnings. Extract it as a module-level private class _ProgressFile.

src/aleph_client/commands/instance/__init__.py (line 1161): Minor inconsistency: reinstall uses Prompt.ask(...) (rich) while all backup.py commands use typer.prompt(...) for the same CRN URL prompt. Pick one; typer.prompt is used everywhere else in the codebase.

src/aleph_client/commands/instance/backup.py (line 1): No tests added. None of the five new commands (create, info, download, delete, restore) or reinstall have test coverage. At minimum, unit tests for the argument-validation logic (mutual-exclusion checks, file-not-found, domain resolution fallback) and mocked happy-path tests for each command would significantly reduce regression risk for these destructive operations.