deps: bump substrate to 961883a8 (imagecache Phase 1, root-cures atelet OOM) - #11
Closed
feifeigood wants to merge 2 commits into
Closed
deps: bump substrate to 961883a8 (imagecache Phase 1, root-cures atelet OOM)#11feifeigood wants to merge 2 commits into
feifeigood wants to merge 2 commits into
Conversation
…ed actor API Bump github.com/agent-substrate/substrate from 3cb7433bd8a8 to 961883a86749 (agent-substrate/substrate#467, on-disk layer-deduplicated image cache + overlayfs rootfs; memorypullcache removed). Root-cures the atelet OOM cycle (substrate#437 / feifeigood/agentfleet#294) and drops oci_unpack from 15-20s to single-digit ms on resume. The bump is wire-breaking: actors and atespaces are now addressed by name-based ObjectRef/ResourceMetadata instead of ActorRef/ActorId, and CreateActor/CreateAtespace return the resource directly. Adapt the three consumers accordingly: - internal/k8s/ate/client.go: wrap create requests in Actor/Atespace with ResourceMetadata; ObjectRef for resume/suspend; CreateActor now returns *ateapipb.Actor. - internal/harness/substrate/substrate.go: GetActorId -> Metadata.Name in the authoritative-resume identity check. - internal/harness/harnesstest/harnesstest.go: mock Control server updated to the new RPC signatures. Plan: feifeigood/agentfleet#369 (P0). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017TK34TZokEcfPq5JdbUVWo
Second bump within #369 P1 soak window. Absorbs five fixes on top of imagecache Phase 1 (961883a8), stopping deliberately before the 07-24 afternoon batch (mTLS google#237, external volumes #405) which has zero soak: - #498 async worker release uses background ctx (zombie-worker fix) - #508 workflow_pause crash when pod is gone - #499 k8sjwt: skip unusable keys in JWKS discovery - #505 valkey cert reload before expiry - #510 redis not-found error fix (also carries #482 Lock API auto-renew rework, in between) No ax code changes needed: none of these touch the ate client API. go build / go vet / go test -race ./... all green (13 packages). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017TK34TZokEcfPq5JdbUVWo
Owner
Author
|
Superseded by #12: reworked to follow all upstream (07-22 refactor) first via merge, then stack the substrate bump as a single squashed patch commit. |
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
1 participant
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.
What
Bump
github.com/agent-substrate/substrate3cb7433bd8a8→961883a86749(merge commit of substrate#467, Phase 1 of substrate#463): on-disk content-addressed layer-deduplicated image cache + overlayfs actor rootfs,memorypullcacheremoved entirely.Plan & context: feifeigood/agentfleet#369 (this is P0). Root-cures the atelet OOM cycle (substrate#437 / feifeigood/agentfleet#294); upstream measured
oci_unpack15–20s → 2.9–4.8ms, cache survives atelet restarts, restarts no longer strand actors.Wire-breaking adaptation
The 66-commit drift renames actor addressing from
ActorRef/ActorIdto name-basedObjectRef/ResourceMetadata, andCreateActor/CreateAtespacenow return the resource directly. Exactly three consumers in this repo, all mechanical:internal/k8s/ate/client.go— request/response shapes (Actor/AtespacewrapResourceMetadata;ObjectReffor resume/suspend;CreateActorreturns*ateapipb.Actor)internal/harness/substrate/substrate.go—GetActorId()→GetMetadata().GetName()in the authoritative-resume identity checkinternal/harness/harnesstest/harnesstest.go— mock Control server updated to the new RPC signaturesTransitive bumps pulled in by
go mod tidy:x/syncv0.21.0,x/sysv0.46.0,x/textv0.39.0.Verification
go build ./...+go vet ./...cleango test -race ./...— 13 packages ok, 0 failures (includes the warm shutdown-drain suite from fix(substrate): drain warm actors on shutdown to prevent RUNNING leak #5, already in base)Deployment notes (do NOT merge-and-deploy casually)
961883a8together with this ax build; rollback must also be done as a pair.🤖 Generated with Claude Code
https://claude.ai/code/session_017TK34TZokEcfPq5JdbUVWo