Skip to content

sync: re-pin giantswarm onto v0.5.0-alpha2 - #346

Merged
heraldbot[bot] merged 0 commit into
giantswarmfrom
sync/20261012-v0.5.0-alpha2
Oct 12, 2026
Merged

heraldbot[bot] merged 0 commit into
giantswarmfrom
sync/20261012-v0.5.0-alpha2

Conversation

@heraldbot

@heraldbot heraldbot Bot commented Oct 12, 2026 •

Copy link
Copy Markdown

Re-pin of giantswarm from v0.4.0-alpha1 onto v0.5.0-alpha2, finished by hand on this branch (description written by an agent on the maintainer's behalf).

Problem

Upstream kagent's go.mod names Substrate v0.5.0-alpha2; the line is still on v0.4.0-alpha1, so the kagent line cannot re-pin onto upstream kagent main.

Proposed solution

git rebase --onto v0.5.0-alpha2 v0.4.0-alpha1 of the line at fe803596, the ledger-only commits folded into one ledger commit, then the line's newer commit 926eedfe cherry-picked. Dropped because upstream carries them: the cherry-picks of upstream agent-substrate#2110 and agent-substrate#2111, the atelet refusal of an empty trust-bundle projection (upstream 601c03ba), the identity e2e trust-bundle rotation, the serverboot half of the OTEL_TRACES_EXPORTER=none patch (upstream agent-substrate#2193). FORK.md and fork/ list every drop and adaptation.

Persisted formats this step changes, and their migration

Row v1.8.x wrote v0.5.0-alpha2 reads Migration
Actor LocalSnapshot field 2 node_vms_with_local_snapshots (list), field 3 content_scope field 2 content_scope; the node in node_name (field 5, the line's) on read: node from the old list, scope from field 3
Actor ActorStatus node only on the worker assignment and the local snapshot assigned_node (field 10) decides where the actor lives on read: the worker's node, else the local snapshot's
ActorTemplate SnapshotConfig on_pause (field 1) gone; a pause captures on_commit (upstream agent-substrate#2309) none needed: each local snapshot records its own scope
PostgreSQL schema migrations 1–3 and the line's own unchanged none
gVisor snapshot layout PerVolume unchanged (upstream agent-substrate#2283 is micro-VM only) none

The migration is the line's own (fix(ateapi): read a 1.8 actor's local-snapshot node and assigned node): upstream keeps no compatibility before v1.0.0. A migrated row is persisted by the actor's next update.

Wire changes between ateom, atelet and ate-api

  • SetWorkerCapacity is RegisterWorker on WorkerService and AteomSupport, with a HardwareIdentity (upstream Worker: Report node HardwareIdentity during Worker registration agent-substrate/substrate#2282). During the roll an ateom of the old release retries its capacity report until its pool rolls; workers it registered keep their capacity, and the scheduler does not read the hardware identity.
  • MintActorJWTRequest.actor_uid is gone (upstream MintActorJWT no longer takes the actor's UID agent-substrate/substrate#2207); the egress gateway's field is ignored.
  • atelet ExternalVolumeSource.access_mode and read_only move to fields 10001 and 10002 (upstream took field 4 for publish_context); across the roll an old ate-api's mode reads as the default read-write-once.
  • The control API answers through internal/apierror: every carried status.Errorf or status.Code on its errors is ported, or a carried Unavailable would reach the caller as Internal.
  • The ateom proto is unchanged; EgressPolicy and EgressRule are unchanged, so the egress-policy contract stays v0.4.0-alpha1.

Chart

The line keeps its own PostgreSQL interface and the in-chart certificate controller (upstream moved both out), so an installation upgrades in place; upstream's per-component resources defaults (with memory limits) are taken. The data plane stays on the agentgateway line's 2.3.0.

Acceptance criteria

  • go build ./..., go vet ./..., go test ./..., helm unittest charts/substrate (100/100) and the gofmt, boilerplate, go-modules, codegen, crd-chart and golangci-lint verifiers pass locally on 79469558
  • run-tests, govulncheck and fork-ledger pass on this branch's head
  • Actors created on a v1.8.x release resume, suspend and run a turn after an in-place upgrade to the release candidate of this line
  • Landed with Actions → sync-upstream, land = sync/20261012-v0.5.0-alpha2, line = giantswarm; never merged

Line head: 926eedf

@heraldbot
heraldbot Bot requested a review from a team as a code owner October 12, 2026 16:37
@teemow teemow changed the title sync: re-pin giantswarm onto v0.5.0-alpha2 (hand-over: a carried patch conflicts) sync: re-pin giantswarm onto v0.5.0-alpha2 Oct 12, 2026
@heraldbot

heraldbot Bot commented Oct 12, 2026

Copy link
Copy Markdown
Author

Landed: giantswarm is now 73da0a6de8b777afbdfadfc07417619cfd9714ba, the head the required checks passed on.

@heraldbot
heraldbot Bot merged commit 73da0a6 into giantswarm Oct 12, 2026
16 checks passed
@heraldbot

heraldbot Bot commented Oct 12, 2026

Copy link
Copy Markdown
Author

Landed: giantswarm is now 73da0a6de8b777afbdfadfc07417619cfd9714ba, the head the required checks passed on.

@heraldbot
heraldbot Bot deleted the sync/20261012-v0.5.0-alpha2 branch October 12, 2026 19:48
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

0 participants