Repository navigation
Release cortex 0.7.3 - #1459
Merged
Merged
Release cortex 0.7.3#1459
Conversation
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
cortex-ai-agents
Bot
requested review from
PhilippMatthes,
SoWieMarkus,
auhlig,
juliusclausnitzer,
mblos and
umswmayj
as code owners
November 2, 2026 12:23
Merged
Contributor
Test Coverage ReportTest Coverage 📊: 71.8% |
umswmayj
approved these changes
Nov 2, 2026
umswmayj
added a commit
that referenced
this pull request
Nov 2, 2026
## Release cortex 0.7.3 ## Release PR [#1458](#1458) — 2026-11-02 ### Bumped Versions | Package | Previous | New | |---|---|---| | cortex | 0.7.2 | 0.7.3 | | cortex-postgres | 0.6.29 | 0.6.30 | | cortex-shim | 0.1.34 | 0.1.35 | | bundles | 0.0.109 | 0.0.110 | | cortex-placement-shim | 0.1.34 | 0.1.35 | ###⚠️ Deprecation - **Slot-based failover reservations deprecated:** The older slot-based failover reservation controller is now deprecated in favor of the new whole-host failover reservation controller (see Features below). The slot-based controller is retained only for migration. New deployments should enable the failover host reservations controller. ### Features - **Failover host controller** ([#1417](#1417)) — A brand-new failover host controller replaces the slot-based failover system with whole-host reservations. Instead of reserving scattered capacity slots, Cortex now keeps a fraction of entire hosts empty so that when a host dies its full workload can be gang-evacuated onto a single reserved host in one move. Hosts are automatically grouped into **failover partitions** — sets of interchangeable hosts discovered from the scheduler's own placement decisions (per flavor), so partitions always reflect live filters, weights, and constraints without manual configuration. Includes a new `failover_partition_graph` KPI plugin with full test coverage, and extensive documentation (`docs/03-reservations-and-inventory/04-failover-host-reservations.md`). Six new alerts added: `CortexNovaFailoverHostCalculationStalled`, `CortexNovaFailoverHostCalculationAborting`, `CortexNovaFailoverHostReconcileErrors`, `CortexNovaFailoverHostSchedulerErrors`, and supporting metrics (`cortex_failover_host_calculation_runs_total`, `cortex_failover_host_calculation_aborts_total`, `cortex_failover_host_reservations_errors_total`, `cortex_failover_host_scheduler_requests_total`). - **`CommittedResourceChange` CRD & reconciler** ([#1432](#1432), [#1446](#1446)) — Introduces the `CommittedResourceChange` (CRC) custom resource definition and a reconciler happy path, enabling a transactional architecture for committed-resource lifecycle management. The API handler or Limes syncer creates a CRC; the CRC reconciler drives its FSM, writes `CommittedResource` specs, and rolls back on failure — no controller writes its own spec (spec flows down, status flows up). Includes new CRD types (`committed_resource_change_types.go`), a Helm CRD manifest (`cortex.cloud_committedresourcechanges.yaml`), RBAC cluster-role updates for `committedresourcechanges`, and architecture documentation (`03-b-committed-resource-transaction-architecture.md`). - **`index_zz` self-registration pattern** ([#1450](#1450)) — Replaces hand-maintained plugin registration maps with a self-registering `zz_index.go` pattern across all cortex plugin families (datasources, extractors, KPIs). The old `supported_syncers.go`, `supported_extractors.go`, and `supported_kpis.go` files have been removed. Adding a new plugin no longer requires editing a central index file. - **Reservations monitoring moved to KPI layer** ([#1445](#1445)) — Reservation state monitoring has been lifted out of the quota-enforcement scheduling filter and into the KPI subsystem (`reservation_state.go`), decoupling observability from the scheduling hot path. The old `filter_quota_enforcement_metrics.go` (111 lines) was removed. - **Pipeline monitor refactor** — Replaced the singleton pipeline monitor with an event-based reporting mechanism, improving observability and reducing coupling. ### Bug Fixes - **Log metrics now include info level** ([#1433](#1433)) — Ensures informational log entries are captured in metrics; previously only warn/error levels were counted. - **Avoid race conditions between agent workflows** ([#1449](#1449)) — Serialized the review workflow to prevent concurrent agent runs from colliding. ### Chores & Refactoring - **Enhanced `make` command-line output** ([#1447](#1447)) — Cleaner, more readable build output. - **Removed redundant lint phase from `make` target** ([#1451](#1451)) — Streamlines the build by eliminating a duplicate lint step. - **Enforce safe context in Tiltfile** ([#1448](#1448)) — Guards local development against accidentally running against the wrong cluster. Configure allowed contexts via the `TILT_SAFE_CONTEXTS` environment variable. ### Documentation - Updated documentation for make targets, phase runner, tool versions ([#1453](#1453)), Tiltfile safe-context guard, and the new plugin self-registration pattern. ### Dependency Updates | Dependency | Old | New | Notes | |---|---|---|---| | `kube-prometheus-stack` | v92.14.0 | **v92.17.0** | via v92.14.1, v92.16.0, v92.16.1 | | `golang.org/x/net` | — | **v0.62.0** | 🔒 **Security patch** | | `debian:trixie-slim` | — | Latest digest | 2 digest updates | | `github.com/sapcc/go-bits` | — | Latest digest | 2 digest updates | | External dependencies | — | **v0.37.2** | | ###⚠️ Operational Notes 1. **CRD installation ordering:** The new `CommittedResourceChange` CRD must be installed before the manager is upgraded. Helm handles this via the `crds/` folder, but verify ordering in CI/CD pipelines. 2. **RBAC changes:** The manager's cluster role now includes `committedresourcechanges`. Ensure RBAC manifests are applied before the new controller starts. 3. **Security dependency — prioritize rollout:** `golang.org/x/net v0.62.0` patches a security vulnerability. This release should be prioritized accordingly. 4. **Metrics relocation — update dashboards/alerts:** Reservation state metrics have moved from the scheduling filter (`filter_quota_enforcement_metrics`) to the KPI subsystem (`reservation_state`). Dashboards or alerts referencing the old metric names must be updated. New alert rules have been added in `helm/bundles/cortex-nova/templates/alerts.yaml`. 5. **Failover host alerts:** Six new Prometheus alert rules have been added for the failover host controller. Review thresholds and routing before rollout. **Contributors:** @umswmayj, @PhilippMatthes, @SoWieMarkus, @mblos, @renovate[bot], @cortex-ai-agents[bot] **PR:** [#1458](#1458) | **Commits:** 27 | **Files changed:** 100 | **+2,827 / −464** ## Dependencies - Prep PR: #1459 (must be merged before this PR)
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.
Summary
Motivation
Release prep for #1458: changelog entry and helm chart version bumps. Merge this before merging #1458.
Test plan
🤖 Generated with Claude Code