Repository navigation
fix(configurator): clean stale runners and unblock additive IAM drafts - #176
Merged
Merged
Conversation
lweberru
requested review from
Breeless,
JoshSTACKIT,
LembDo and
dweezl
as code owners
October 28, 2026 12:48
lweberru
force-pushed
the
fix/runner-retention-and-permission-additions
branch
from
October 28, 2026 12:50
314c1e4 to
b1899ff
Compare
lweberru
added a commit
that referenced
this pull request
Oct 28, 2026
A recovered warm Runner could remain desired STARTED/CRASHED indefinitely because normal pool retirement only accepted succeeded/failed dispatches, while a reconciled interruption retains reconciliation_required as historical status. Retirement now also accepts the existing validated recovery-replan clearance after Runner tickets expire and with no active guided recovery. Exact space/origin/job/app/package bindings remain mandatory. The stopped-app sweeper uses the same established validator instead of checking only receipt existence, preserving protection when a later separate recovery State invalidates clearance. Historical migrations are unchanged; the additive migration is 077. Validation: 69 PostgreSQL identity, isolation, recovery and retention tests pass. The actual cleared-apply fixture tests expired authority positively and live tickets/missing clearance negatively, with rollback preserving later recovery assertions. Independent review found no remaining issues. English/German operations documentation is updated. This follow-up changes no Assistant implementation, knowledge, intent schema or qualification oracle; the Assistant qualification for parent #176 remains separately required before the combined Dev deployment. No production or customer Plan/Apply/State/lock mutation is included.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stopped Runner apps could remain in CF after failed cleanup or creation interrupted before durable binding. A bounded supplementary sweep now examines 50 apps per minute and removes stopped execution apps aged at least 24 hours only after authoritative job/ticket/recovery checks and final CF identity, route, binding, task and process checks. Associated completed apps are eligible; unassociated apps additionally need this Configurator origin’s ownership label, set on all newly created runners. Unknown unlabelled legacy apps and unresolved recovery remain protected and appear as counts in sanitized maintenance summaries. Only confirmed CF absence creates a cleanup receipt. Job history, State and locks are unchanged.
The Assistant now proposes requested service permissions on an existing granular template even when its owner field is unfinished. It preserves that field, all existing rights and the unrequested recipient list. New-profile and publication owner requirements remain enforced. The exact Public/Kubernetes request is covered with both empty and configured owner values.
Validation: full local checks pass (1,382 tests; 76 expected skips), real PostgreSQL identity/recovery/retention tests pass (69), and all 34 affected real-model template checks pass across two repetitions. Independent review covered ownership, active jobs, Recovery, pagination and publication guards; its asynchronous-DELETE receipt finding was fixed and covered by 202→404 and 202→still-present regressions. English/German operational and Assistant documentation is updated. Dev deployment is authorized separately in the current scope; no production or customer Plan/Apply/State mutation is included.