amadan

carlos / platform Public

Merge remote-tracking branch 'origin/main' into bootstrap-carlos-follows-316

# Conflicts:
#	internal/provision/plan14_root_reads_test.go
claude a84a1962d6f6ed6fc9454b467882ea96ad70827b parent ce50cbbparent d7a5b8e
7 files changed, +360 −6
  • internal/host/agent.go +30 −6
  • internal/host/rootunits_test.go +88 −0
  • internal/provision/plan14_root_reads_test.go +9 −0
  • internal/provision/testdata/plan14-root-reads-adopt-unit-root-pass-315.tsv +7 −0
  • internal/release/adopt.go +48 −0
  • internal/release/rollback.go +26 −0
  • internal/release/rootunits_test.go +152 −0

CI passed — run details

CI log 
=== gate ===
go: downloading github.com/aws/aws-sdk-go-v2/service/autoscaling v1.74.0
go: downloading github.com/aws/aws-sdk-go-v2/service/ssm v1.74.0
cd launchctx && go vet ./... && go test ./... -count=1
ok  	amadan.net/carlos/platform.git/launchctx	0.012s
?   	amadan.net/carlos/platform.git/launchctx/controlname	[no test files]
?   	amadan.net/carlos/platform.git/launchctx/wireid	[no test files]
go vet ./...
gofmt -l .: clean
go test ./... -count=1 -timeout 30m
ok  	github.com/carlosframework/platform	315.840s
?   	github.com/carlosframework/platform/cmd/changelog	[no test files]
ok  	github.com/carlosframework/platform/cmd/console	2.743s
ok  	github.com/carlosframework/platform/cmd/moveprobe	93.325s
ok  	github.com/carlosframework/platform/cmd/zddprobe	0.710s
ok  	github.com/carlosframework/platform/internal/activator	250.679s
ok  	github.com/carlosframework/platform/internal/agentsock	0.016s
ok  	github.com/carlosframework/platform/internal/alerting	0.004s
ok  	github.com/carlosframework/platform/internal/analytics	0.019s
ok  	github.com/carlosframework/platform/internal/appconfig	2.471s
ok  	github.com/carlosframework/platform/internal/apperr	0.089s
ok  	github.com/carlosframework/platform/internal/applog	0.160s
ok  	github.com/carlosframework/platform/internal/awsinv	0.004s
ok  	github.com/carlosframework/platform/internal/boxchoice	0.004s
ok  	github.com/carlosframework/platform/internal/boxid	0.783s
?   	github.com/carlosframework/platform/internal/boxid/boxidtest	[no test files]
?   	github.com/carlosframework/platform/internal/boxid/internal/signtest	[no test files]
ok  	github.com/carlosframework/platform/internal/boxvolumes	0.011s
ok  	github.com/carlosframework/platform/internal/caps	0.012s
ok  	github.com/carlosframework/platform/internal/certwarm	0.002s
ok  	github.com/carlosframework/platform/internal/cgroup	0.707s
ok  	github.com/carlosframework/platform/internal/changelog	0.316s
ok  	github.com/carlosframework/platform/internal/claimactivation	0.803s
ok  	github.com/carlosframework/platform/internal/clamd	0.064s
?   	github.com/carlosframework/platform/internal/cliflag	[no test files]
ok  	github.com/carlosframework/platform/internal/console	408.413s
?   	github.com/carlosframework/platform/internal/controlname	[no test files]
?   	github.com/carlosframework/platform/internal/controlsock	[no test files]
ok  	github.com/carlosframework/platform/internal/dbdoor	6.081s
ok  	github.com/carlosframework/platform/internal/dbdoor/dbdoortest	0.010s
ok  	github.com/carlosframework/platform/internal/dbverify	0.005s
ok  	github.com/carlosframework/platform/internal/dnsrec	0.003s
ok  	github.com/carlosframework/platform/internal/docsite	0.005s
ok  	github.com/carlosframework/platform/internal/domains	2.369s
ok  	github.com/carlosframework/platform/internal/doorkey	0.006s
ok  	github.com/carlosframework/platform/internal/economics	0.004s
ok  	github.com/carlosframework/platform/internal/edge	45.123s
?   	github.com/carlosframework/platform/internal/edgeproof	[no test files]
ok  	github.com/carlosframework/platform/internal/edgesfile	0.009s
ok  	github.com/carlosframework/platform/internal/fleetapi	0.132s
ok  	github.com/carlosframework/platform/internal/gate	1.655s
ok  	github.com/carlosframework/platform/internal/genselect	0.016s
ok  	github.com/carlosframework/platform/internal/geo	0.015s
ok  	github.com/carlosframework/platform/internal/geoip	0.032s
ok  	github.com/carlosframework/platform/internal/historyreadback	0.153s
ok  	github.com/carlosframework/platform/internal/host	181.314s
ok  	github.com/carlosframework/platform/internal/hostactors	6.206s
ok  	github.com/carlosframework/platform/internal/hostactors/awsmembership	0.015s
ok  	github.com/carlosframework/platform/internal/hostbox	2.777s
ok  	github.com/carlosframework/platform/internal/hostclaim	0.572s
ok  	github.com/carlosframework/platform/internal/hostenv	0.015s
ok  	github.com/carlosframework/platform/internal/hostevidence	0.015s
ok  	github.com/carlosframework/platform/internal/hostexec	5.336s
ok  	github.com/carlosframework/platform/internal/hostfence	0.040s
ok  	github.com/carlosframework/platform/internal/hostfiles	10.653s
ok  	github.com/carlosframework/platform/internal/hostfloor	0.015s
ok  	github.com/carlosframework/platform/internal/hostguard	0.995s
ok  	github.com/carlosframework/platform/internal/hostops	1.604s
?   	github.com/carlosframework/platform/internal/hostops/fencerelease	[no test files]
ok  	github.com/carlosframework/platform/internal/hostops/internal/fencerelease	0.020s
ok  	github.com/carlosframework/platform/internal/hostquarantine	0.015s
ok  	github.com/carlosframework/platform/internal/hostrelocate	0.427s
ok  	github.com/carlosframework/platform/internal/hostresolve	0.156s
ok  	github.com/carlosframework/platform/internal/hostwrite	0.014s
ok  	github.com/carlosframework/platform/internal/infraview	0.016s
ok  	github.com/carlosframework/platform/internal/instancectl	5.256s
ok  	github.com/carlosframework/platform/internal/instanceexec	2.754s
ok  	github.com/carlosframework/platform/internal/instancefs	0.020s
ok  	github.com/carlosframework/platform/internal/instancekey	0.015s
ok  	github.com/carlosframework/platform/internal/instanceupdates	0.022s
ok  	github.com/carlosframework/platform/internal/layout	1.573s
ok  	github.com/carlosframework/platform/internal/lbhealth	0.002s
ok  	github.com/carlosframework/platform/internal/lease	9.815s
ok  	github.com/carlosframework/platform/internal/lease/leasetest	0.006s
ok  	github.com/carlosframework/platform/internal/ledger	0.016s
ok  	github.com/carlosframework/platform/internal/mail	0.171s
ok  	github.com/carlosframework/platform/internal/maintenance	0.077s
ok  	github.com/carlosframework/platform/internal/maintenanceexec	5.540s
ok  	github.com/carlosframework/platform/internal/maintenanceguard	0.081s
ok  	github.com/carlosframework/platform/internal/metrics	0.005s
ok  	github.com/carlosframework/platform/internal/moves	21.165s
ok  	github.com/carlosframework/platform/internal/moves/archive	0.794s
ok  	github.com/carlosframework/platform/internal/objstore	0.186s
ok  	github.com/carlosframework/platform/internal/ops	0.175s
ok  	github.com/carlosframework/platform/internal/park	0.203s
ok  	github.com/carlosframework/platform/internal/privatefile	0.005s
ok  	github.com/carlosframework/platform/internal/provision	38.193s
ok  	github.com/carlosframework/platform/internal/publicrelease	0.012s
?   	github.com/carlosframework/platform/internal/regionname	[no test files]
ok  	github.com/carlosframework/platform/internal/registry	5.538s
ok  	github.com/carlosframework/platform/internal/relay	7.254s
ok  	github.com/carlosframework/platform/internal/release	24.591s
ok  	github.com/carlosframework/platform/internal/rollout	3.961s
ok  	github.com/carlosframework/platform/internal/rolloutwords	0.003s
ok  	github.com/carlosframework/platform/internal/rootaws	0.007s
ok  	github.com/carlosframework/platform/internal/rootparticipation	0.058s
ok  	github.com/carlosframework/platform/internal/rootstate	0.026s
ok  	github.com/carlosframework/platform/internal/runs	0.572s
ok  	github.com/carlosframework/platform/internal/s3compat	0.005s
ok  	github.com/carlosframework/platform/internal/safefile	0.004s
ok  	github.com/carlosframework/platform/internal/schedule	2.543s
ok  	github.com/carlosframework/platform/internal/seeds	0.011s
ok  	github.com/carlosframework/platform/internal/sim/hostlife	10.341s
ok  	github.com/carlosframework/platform/internal/sim/litestreamfacts	3.558s
ok  	github.com/carlosframework/platform/internal/status	0.061s
ok  	github.com/carlosframework/platform/internal/steering	0.006s
ok  	github.com/carlosframework/platform/internal/stopevidence	0.221s
ok  	github.com/carlosframework/platform/internal/svcrole	0.002s
ok  	github.com/carlosframework/platform/internal/telemetry	0.004s
ok  	github.com/carlosframework/platform/internal/termfence	0.004s
ok  	github.com/carlosframework/platform/internal/tunnelclient	0.004s
ok  	github.com/carlosframework/platform/internal/unixdial	0.003s
ok  	github.com/carlosframework/platform/internal/updatewords	0.008s
ok  	github.com/carlosframework/platform/internal/usage	0.005s
ok  	github.com/carlosframework/platform/internal/version	0.002s
ok  	github.com/carlosframework/platform/internal/vet	1.590s
=== cross ===
--- shard 0 of 5 ---
cross-compiling darwin/arm64
--- shard 1 of 5 ---
cross-compiling darwin/amd64
--- shard 2 of 5 ---
cross-compiling linux/arm64
linux/arm64 carlos binary: 49414306 bytes (ceiling 52000000)
--- shard 3 of 5 ---
cross-compiling windows/amd64
--- shard 4 of 5 ---
cross-compiling windows/arm64
=== minio-contract ===
minio contract: arm64, minio RELEASE.2025-09-07T16-13-09Z, mc RELEASE.2025-08-13T08-35-41Z
  minio: downloading https://github.com/minio/minio/releases/download/RELEASE.2025-09-07T16-13-09Z/minio.linux-arm64.RELEASE.2025-09-07T16-13-09Z
  minio: downloaded, digest verified
  mc: downloading https://github.com/minio/mc/releases/download/RELEASE.2025-08-13T08-35-41Z/mc.linux-arm64.RELEASE.2025-08-13T08-35-41Z
  mc: downloaded, digest verified
  minio healthy on 127.0.0.1:20999
Added `local` successfully.
Bucket created successfully `local/contract-bucket`.
Bucket created successfully `local/denied-bucket`.
Created policy `contract-only` successfully.
Added user `scopeduser` successfully.
Attached Policies: [contract-only]
To User: scopeduser
  buckets and scoped user ready
=== RUN   TestDeliveredContractMinio
=== RUN   TestDeliveredContractMinio/put
=== RUN   TestDeliveredContractMinio/get_round-trips_the_body
=== RUN   TestDeliveredContractMinio/list_shows_the_key
=== RUN   TestDeliveredContractMinio/head_and_list_agree_on_the_version
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_its_own_object
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_its_own_object/not_by_PutObjectTagging
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_its_own_object/and_not_by_x-amz-tagging_on_upload
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/not_by_CreateMultipartUpload
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/not_by_CopyObject_with_a_REPLACE_tagging_directive
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/but_multipart_itself_is_allowed
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/and_copy_itself_is_allowed
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/a_POST-policy_forged_tag_with_a_matching_version_is_still_not_a_verdict
=== RUN   TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/and_an_ordinary_tagged-with-something-else_upload_is_allowed
=== RUN   TestDeliveredContractMinio/delete_removes_it
=== RUN   TestDeliveredContractMinio/cross-bucket_denial
--- PASS: TestDeliveredContractMinio (0.02s)
    --- PASS: TestDeliveredContractMinio/put (0.00s)
    --- PASS: TestDeliveredContractMinio/get_round-trips_the_body (0.00s)
    --- PASS: TestDeliveredContractMinio/list_shows_the_key (0.00s)
    --- PASS: TestDeliveredContractMinio/head_and_list_agree_on_the_version (0.00s)
    --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_its_own_object (0.00s)
        --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_its_own_object/not_by_PutObjectTagging (0.00s)
        --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_its_own_object/and_not_by_x-amz-tagging_on_upload (0.00s)
    --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route (0.01s)
        --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/not_by_CreateMultipartUpload (0.00s)
        --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/not_by_CopyObject_with_a_REPLACE_tagging_directive (0.00s)
        --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/but_multipart_itself_is_allowed (0.00s)
        --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/and_copy_itself_is_allowed (0.00s)
        --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/a_POST-policy_forged_tag_with_a_matching_version_is_still_not_a_verdict (0.00s)
        --- PASS: TestDeliveredContractMinio/the_app_cannot_tag_by_any_other_route/and_an_ordinary_tagged-with-something-else_upload_is_allowed (0.00s)
    --- PASS: TestDeliveredContractMinio/delete_removes_it (0.00s)
    --- PASS: TestDeliveredContractMinio/cross-bucket_denial (0.00s)
PASS
ok  	github.com/carlosframework/platform/internal/objstore	0.024s
=== RUN   TestT4StorageProbeMinio
2026/10/20 00:25:47 moves: T4: STORE: list: operation error S3: ListObjectsV2, https response error StatusCode: 403, RequestID: 18E0152FFDC110F0, HostID: 7c420053e0fbcd7f41653fdd92ffa2802c903d05303b1aa15adc2e8bb65d183f, api error SignatureDoesNotMatch: The request signature we calculated does not match the signature you provided. Check your key and signing method.
--- PASS: TestT4StorageProbeMinio (0.01s)
PASS
ok  	github.com/carlosframework/platform/internal/host	0.021s
minio contract: PASS
diff --git a/internal/host/agent.go b/internal/host/agent.go
index 5ccaa3955..bb9e33fea 100644
--- a/internal/host/agent.go
+++ b/internal/host/agent.go
@@ -1295,14 +1295,19 @@ func adoptLoop(reg *registry.Registry, svc *activator.Service, sd *scheduleDeps,
// DO NOT add the platform self-update here either: the system leg
// lives in the fast loop and re-observes hourly on its own, so an
// hourly call here would only double the pointer read.
- if err := release.Pass(ctx, reg, store, pub, optDir, restart,
+ //
+ // agentUnitsOption hands named units to root when this process is
+ // not root (issue #315), which leaves WithStartUnit and
+ // WithRollbackRestart to an agent run as root.
+ opts := append([]release.PassOption{
release.BakeDir(dataDir()), release.OnStageBlocked(stageRequester(dataDir())),
release.RestartHibernating(restart),
release.WithStartUnit(makeStartRoute(reg, systemctlStartUnit)),
release.WithRollbackRestart(agentRollbackUnitRestart(reg, systemdUnits{}, optDir, os.Getenv)),
release.WithProofGate(agentProofGate()), release.WithHostWrites(writes),
release.WithDigestRatchet(pointerRatchet(false, optDir, dataDir(), os.Getenv)),
- updatesPassOption(dataDir(), false)); err != nil {
+ updatesPassOption(dataDir(), false)}, agentUnitsOption()...)
+ if err := release.Pass(ctx, reg, store, pub, optDir, restart, opts...); err != nil {
log.Printf("agent: adopt pass: %v", err)
}
// The schedule converge's hourly floor (schedule spec §6.2), AFTER the
@@ -1727,9 +1732,10 @@ func stageMarkerRequester(dd string) func(host, ver string, err error) {
}
// One line for every cause the marker relays (Task 12a): a build
// this process cannot stage, an install request only root
- // resolves (ErrInstallRequestNeedsRoot), and a held wake's target
- // only root installs. The throttle does not carry the cause.
- log.Printf("agent: %s needs a root pass for %s (a build only root can stage or install, or an install request only root resolves); asked for one", host, ver)
+ // resolves (ErrInstallRequestNeedsRoot), a named unit only root
+ // restarts (ErrUnitRestartNeedsRoot), and a held wake's target only
+ // root installs. The throttle does not carry the cause.
+ log.Printf("agent: %s needs a root pass for %s (a build only root can stage or install, a unit only root can restart, or an install request only root resolves); asked for one", host, ver)
})
}
@@ -2020,8 +2026,11 @@ type restartFailure struct {
// did not already need, and it cannot fire for a sleeper: only an
// ERRORED restart reaches it, and every arm that can meet a sleeping
// route swallows its failure.
+//
+// unitsOption hands named units to root when this process is not root
+// (issue #315); see agentUnitsOption.
func fastPassOpts(dd string, stageReq func(host, ver string, err error), restart, start func(string) error) []release.PassOption {
- return []release.PassOption{
+ return append([]release.PassOption{
release.BakeDir(dd),
release.OnStageBlocked(stageReq),
release.RestartHibernating(restart),
@@ -2031,7 +2040,22 @@ func fastPassOpts(dd string, stageReq func(host, ver string, err error), restart
// resolves. The book is opened per pass, not here: these options are
// built once per agent run.
updatesPassOption(dd, false),
+ }, agentUnitsOption()...)
+}
+
+// agentRunsAsRoot is the agent's euid test, a seam for tests.
+var agentRunsAsRoot = func() bool { return os.Geteuid() == 0 }
+
+// agentUnitsOption is release.RootRestartsUnits for an agent that is not
+// root (issue #315). carlos-edge.service runs as User=carlos, and systemd
+// refuses that user a restart of any unit a route names, so a named-unit
+// route goes to the root pass through the stage relay before anything moves.
+// An agent run as root restarts the unit itself, as before.
+func agentUnitsOption() []release.PassOption {
+ if agentRunsAsRoot() {
+ return nil
}
+ return []release.PassOption{release.RootRestartsUnits()}
}
// fastMembers computes the fast set for one tick — the routes with a reason
diff --git a/internal/host/rootunits_test.go b/internal/host/rootunits_test.go
new file mode 100644
index 000000000..6c2df491a
--- /dev/null
+++ b/internal/host/rootunits_test.go
@@ -0,0 +1,88 @@
+package host
+
+import (
+ "context"
+ "errors"
+ "runtime"
+ "testing"
+ "time"
+
+ "github.com/carlosframework/platform/internal/registry"
+ "github.com/carlosframework/platform/internal/release"
+)
+
+// Issue #315, through the fast loop's production options: an agent that is
+// not root hands a named-unit route to root's pass rather than repointing it
+// and failing systemd's permission check. An agent run as root keeps
+// restarting the unit itself.
+func TestFastSweepHandsANamedUnitToRoot(t *testing.T) {
+ for _, tc := range []struct {
+ name string
+ root bool
+ }{{"agent user", false}, {"root", true}} {
+ t.Run(tc.name, func(t *testing.T) {
+ was := agentRunsAsRoot
+ agentRunsAsRoot = func() bool { return tc.root }
+ t.Cleanup(func() { agentRunsAsRoot = was })
+
+ env := newFastFixture(t)
+ ctx := context.Background()
+ plat := runtime.GOOS + "-" + runtime.GOARCH
+ for _, v := range []string{"v1", "v2"} {
+ m := release.Manifest{App: "fastapp", Version: v, Kind: "binary",
+ Artifacts: map[string]release.Artifact{plat: {}}}
+ if err := release.Ship(ctx, env.store, "bab", m, map[string][]byte{plat: []byte("build " + v)}); err != nil {
+ t.Fatalf("Ship %s: %v", v, err)
+ }
+ }
+ if err := release.Promote(ctx, env.store, env.priv, "bab", "fastapp", "canary/dev", "v1"); err != nil {
+ t.Fatal(err)
+ }
+ if err := env.reg.Upsert(registry.Route{Host: "console.test", Kind: "instance", App: "fastapp",
+ Channel: "canary/dev", Status: "ready", Addr: "127.0.0.1:8090"}); err != nil {
+ t.Fatal(err)
+ }
+ if err := release.Pass(ctx, env.reg, env.store, env.pub, env.optDir, nil); err != nil {
+ t.Fatal(err)
+ }
+ if err := env.reg.SetUnit("console.test", "console.service"); err != nil {
+ t.Fatal(err)
+ }
+ if err := release.Promote(ctx, env.store, env.priv, "bab", "fastapp", "canary/dev", "v2"); err != nil {
+ t.Fatal(err)
+ }
+
+ var unitRestarts []string
+ restart := makeRestartRoute(env.reg, func(string) {},
+ unitOnly(func(u string) error { unitRestarts = append(unitRestarts, u); return nil }), nil)
+ var relayed []error
+ stageReq := func(host, ver string, err error) {
+ if host == "console.test" && ver == "v2" {
+ relayed = append(relayed, err)
+ }
+ }
+ d := env.deps(nil)
+ d.restart = restart
+ d.passOpts = fastPassOpts(env.dataDir, stageReq, restart, nil)
+
+ fastTick(ctx, d, time.Unix(1754640000, 0))
+
+ live, err := release.LiveVersion(env.optDir, "console.test")
+ if err != nil {
+ t.Fatal(err)
+ }
+ if tc.root {
+ if live != "v2" || len(unitRestarts) != 1 || len(relayed) != 0 {
+ t.Fatalf("root agent: live %q, unit restarts %v, relays %v; want v2, one restart, no relay", live, unitRestarts, relayed)
+ }
+ return
+ }
+ if live != "v1" || len(unitRestarts) != 0 {
+ t.Fatalf("agent user: live %q, unit restarts %v; want v1 untouched and no restart", live, unitRestarts)
+ }
+ if len(relayed) != 1 || !errors.Is(relayed[0], release.ErrUnitRestartNeedsRoot) {
+ t.Fatalf("relays = %v, want one ErrUnitRestartNeedsRoot", relayed)
+ }
+ })
+ }
+}
diff --git a/internal/provision/plan14_root_reads_test.go b/internal/provision/plan14_root_reads_test.go
index c9c8e9b54..1175ecad9 100644
--- a/internal/provision/plan14_root_reads_test.go
+++ b/internal/provision/plan14_root_reads_test.go
@@ -933,6 +933,15 @@ func TestPlan14RootReadCensus(t *testing.T) {
if err = loadRootReadInventory(filepath.Join(root, "internal/provision/testdata/plan14-root-reads-claim-guard-umask.tsv"), &inv, true); err != nil {
t.Fatal(err)
}
+ // Census overlay adopt-unit-root-pass-315 (amadan #315: a non-root agent
+ // hands a named unit's routes to the root pass), loaded last. PROPOSED-315,
+ // ruling asked of the census owner on #280: adoptHost re-pinned at its
+ // class, type passConfig gains a bool, and unitServed (new) classified
+ // with processGroup, whose membership it inverts. Both envelopes
+ // re-pinned. It adds no accepted-legacy row, so the seal stands.
+ if err = loadRootReadInventory(filepath.Join(root, "internal/provision/testdata/plan14-root-reads-adopt-unit-root-pass-315.tsv"), &inv, true); err != nil {
+ t.Fatal(err)
+ }
// Census overlay bootstrap-carlos-follows-316 (amadan #316: system status
// compares bootstrap/carlos with stable's build), loaded last. Envelope
// only, proposed; ruling asked of the census owner on #280. The seal
diff --git a/internal/provision/testdata/plan14-root-reads-adopt-unit-root-pass-315.tsv b/internal/provision/testdata/plan14-root-reads-adopt-unit-root-pass-315.tsv
new file mode 100644
index 000000000..2844ab497
--- /dev/null
+++ b/internal/provision/testdata/plan14-root-reads-adopt-unit-root-pass-315.tsv
@@ -0,0 +1,7 @@
+# Census overlay adopt-unit-root-pass-315 (amadan #315: a non-root agent hands named-unit routes and their process group to the root pass), loaded LAST, after claim-guard-umask. Proposed; ruling asked of the census owner on #280.
+# adoptHost re-pinned at its class and task: the new hand-off precedes staging and only returns. type passConfig gains a bool. unitServed (new) reads registry routes exactly as processGroup does and classifies with it. carlos adopt (root) never sets RootRestartsUnits, so root reads are unchanged. The seal stands: no accepted-legacy row.
+@coverage internal/release 4aafab144b1b742b27cab13ded61276b5203a64eb94f12203b6b94adddd0ae83
+@coverage internal/host c589ae17ec476291bd5cd103f10909b55ddcacdc1bddff76800078133766665e
+internal/release/adopt.go adoptHost authority B3,E1,E2,E3,E4,J1 E16: root first-registration creation evidence under release exclusion; exact installation identity is required for first-program adoption and is consumed before an attempt. Installed-program history comes from the verified claim launch. Fixtures: TestPlan14FreshInstallRegistration, TestPlan14FreshInstallProgramGate, TestPlan14FreshInstallCanonicalIdentity, TestPlan14FreshInstallLaunchObservation, TestPlan14InstalledLaunchProgramEvidence, TestPlan14FreshInstallProcessCrash, TestPlan14FreshInstallVisibleSync, TestClaimControlUnitInstallers. E22: root generation pins guard existing-tree artifact and manifest writes in adoption and moves-held staging/publication. Held snapshot is read under release exclusion before the leaf held-stage lock; failures remain restrictive for writes. Claims-off source returns no root pins. Fresh wholly absent version trees retain signed staging. Fixtures: TestPlan14GenerationSiblingWrites, TestPlan14GenerationVerifiedSiblingWrites, TestPlan14GenerationHeldStage (production source assembly, both flag values without census, lock order, pin/error/fresh cases). Production completeness remains unavailable; no new completeness grant.;rev3=PROPOSED-p3;run=8;identity=carlos adopt uid 0; legacy agent pass uid carlos;basis=applies root settings to canonically resolved routes including named units while legacy settings eligibility remains unchanged;rev3=PROPOSED-p3;run=16;basis=P3-R11 approved punctuation edit only; authority and reads unchanged;rev3=PROPOSED-315;identity=legacy agent pass uid carlos (RootRestartsUnits; never set on carlos adopt);basis=a non-root agent pass hands a route its named unit serves (unitServed) to root through the stage relay before staging or repointing; root adopt reads unchanged registry:App@11360,registry:App@11520,registry:App@12549,registry:App@13475,registry:App@15054,registry:App@20213,registry:App@25689,registry:App@3144,registry:App@33448,registry:App@6758,registry:App@8059,registry:App@8253,registry:Hibernate@34309,registry:Host@10484,registry:Host@11240,registry:Host@11273,registry:Host@11373,registry:Host@11533,registry:Host@11635,registry:Host@11667,registry:Host@11732,registry:Host@12182,registry:Host@12315,registry:Host@12762,registry:Host@12871,registry:Host@13073,registry:Host@13565,registry:Host@14351,registry:Host@14826,registry:Host@15040,registry:Host@15382,registry:Host@15972,registry:Host@19703,registry:Host@19811,registry:Host@20181,registry:Host@21001,registry:Host@2105,registry:Host@21159,registry:Host@21621,registry:Host@21673,registry:Host@21820,registry:Host@21907,registry:Host@2294,registry:Host@23013,registry:Host@23257,registry:Host@2375,registry:Host@23917,registry:Host@24302,registry:Host@24443,registry:Host@25657,registry:Host@25784,registry:Host@27126,registry:Host@27776,registry:Host@31009,registry:Host@3157,registry:Host@31627,registry:Host@31952,registry:Host@32026,registry:Host@32273,registry:Host@32424,registry:Host@33620,registry:Host@33787,registry:Host@33894,registry:Host@34122,registry:Host@34265,registry:Host@35010,registry:Host@35136,registry:Host@35224,registry:Host@4469,registry:Host@4583,registry:Host@5546,registry:Host@5651,registry:Host@6870,registry:Host@6918,registry:Host@7035,registry:Host@7067,registry:Host@7945,registry:Host@8113,registry:Host@8239,registry:Host@8358,registry:Host@8836,registry:Host@9223,registry:Host@9449,registry:Host@9545,registry:Host@9636,registry:Kind@554,registry:Unit@32172,registry:Unit@579,whole:Route@10073,whole:Route@10427,whole:Route@11355,whole:Route@11515,whole:Route@12539,whole:Route@13465,whole:Route@15022,whole:Route@15587,whole:Route@15948,whole:Route@17171,whole:Route@17474,whole:Route@17630,whole:Route@19960,whole:Route@20104,whole:Route@20203,whole:Route@2206,whole:Route@22742,whole:Route@23141,whole:Route@25679,whole:Route@30580,whole:Route@3139,whole:Route@33596,whole:Route@3595,whole:Route@4347,whole:Route@6753 219694d85c22b7886a4ebc62be97888bf993af79d5c0df134b95e0480b2db992
+internal/release/adopt.go type passConfig authority E2,E3 E19: both destructive readers use the root reservation snapshot under existing release exclusion. Exact artifact and containing-tree identities are retained; missing completeness refuses destruction. Production installation/launch/kernel completeness and verified retirement remain unfinished. Fixtures: TestPlan14GenerationDeletionCannotUnpin, TestPlan14GenerationSnapshotCompleteness, TestPlan14GenerationSnapshotPermitsUnreservedRestage, TestPlan14GenerationSnapshotPinsContainingTree, TestPlan14GenerationSnapshotEmptyIsUnknown, TestPlan14GenerationConsumerMode, TestPlan14GenerationProductionModeGate, TestPlan14GenerationSnapshotPermitsFreshTree.;rev3=PROPOSED-p3;run=8;identity=carlos adopt uid 0; legacy agent pass uid carlos;basis=stores the typed census-only root update provider and per-pass resolved input authority;rev3=PROPOSED-315;identity=legacy agent pass uid carlos (RootRestartsUnits; never set on carlos adopt);basis=adds the unitsToRoot bool set by RootRestartsUnits; reads nothing 60131fde0633bcee3db957f9f898b6e9eef96f1a84f2f2f03a7ab0d9b05db6ad
+internal/release/rollback.go unitServed authority B3,E1,E3,J1 root-route;kept=B3,E1,E3;scope=protected-decision-only;rev3=PROPOSED-315;identity=legacy agent pass uid carlos (RootRestartsUnits; never set on carlos adopt);basis=processGroup membership inverted (same account, app, channel and non-empty upstream as a route naming a unit); only ever narrows the agent pass to a hand-off; a registry list failure hands off; no file read registry:App@281,registry:App@290,registry:Host@244,registry:Host@254,registry:Unit@265,registry:Unit@70,whole:Route@119,whole:Route@233,whole:Route@316,whole:Route@335,whole:Route@351,whole:Route@379,whole:Route@398 fef7b67b13241535f225f634f776d2822837a734ca6a25ea732ef37d9b7909e9
diff --git a/internal/release/adopt.go b/internal/release/adopt.go
index da914a81c..2867b8797 100644
--- a/internal/release/adopt.go
+++ b/internal/release/adopt.go
@@ -86,6 +86,11 @@ type passConfig struct {
// is consulted for them at all.
restartHibernating func(host string) error
+ // unitsToRoot hands every route that names its own unit to the root pass
+ // before anything is staged or repointed (RootRestartsUnits). false, the
+ // default and RunAdopt's, restarts the named unit in this pass.
+ unitsToRoot bool
+
// startUnit is consulted ONLY on adoptHost's restart-failure rollback
// path, after the symlink has gone back (WithStartUnit). nil — every
// call site that does not pass the option, and every existing test —
@@ -154,6 +159,40 @@ type FailedBuild func(acct, app, build, host string, served bool) bool
// this pass sees it. Never shown to members.
var ErrInstallRequestNeedsRoot = errors.New("release: an install request waits for root's pass to resolve it")
+// ErrUnitRestartNeedsRoot is what a RootRestartsUnits pass hands
+// OnStageBlocked's callback for a route that names its own unit: nothing
+// failed, and the version is the build root's pass should install. Never
+// shown to members.
+var ErrUnitRestartNeedsRoot = errors.New("release: a route's named unit is restarted by root's pass")
+
+// RootRestartsUnits is for a pass whose process cannot restart a unit an
+// operator named (issue #315): the agent's in-process loops when they do not
+// run as root. carlos-edge.service is User=carlos, and systemd refuses it
+// ("Interactive authentication required") for console.service and every
+// other route.Unit. Before this option such a pass repointed the live link,
+// failed the restart, read the refusal as a broken build, rolled back,
+// poisoned the version and tried to start the old build over a process that
+// had never stopped.
+//
+// With it, a route that names its unit, and every route in that unit's
+// process group (processGroup: same account, app, channel and upstream), is
+// handed to root before anything is staged or repointed, through
+// OnStageBlocked with ErrUnitRestartNeedsRoot, the way a build this process
+// cannot stage is. The live links keep naming the running build, so no
+// sibling is stamped with a build the unit does not run. Nothing is poisoned
+// and no refusal is recorded; root's
+// pass stages, repoints, restarts and stamps. Every hold, refusal and proof
+// check before the hand-off still applies, so root is only asked for a build
+// this pass would have installed.
+//
+// Lifecycle host writes (WithHostWrites) restart through the selected
+// receiver instead, so the option does nothing under them. Nor for a route
+// this pass would not restart (restart nil: a sleeper on a pass without
+// RestartHibernating, a sidecar).
+func RootRestartsUnits() PassOption {
+ return func(c *passConfig) { c.unitsToRoot = true }
+}
+
// NagExpiredCanaries opts a Pass/PassFilter call into spec §5's nag: one
// line per expired canary/* route the call sees, written to w.
//
@@ -1483,6 +1522,15 @@ func adoptHost(ctx context.Context, work hostwrite.Work, reg *registry.Registry,
if !admitted {
return
}
+ // A named unit this process cannot restart is root's (RootRestartsUnits),
+ // and so is every route its process serves: ask for root's pass and touch
+ // nothing. Silent on the tick; the requester logs each request it writes.
+ if cfg.unitsToRoot && restart != nil && !cfg.hostWrites.Enabled && unitServed(reg, rt) {
+ if cfg.stageBlocked != nil {
+ cfg.stageBlocked(rt.Host, ver, ErrUnitRestartNeedsRoot)
+ }
+ return
+ }
// The evidence that authorizes ver (settings spec §4.1). Only today's
// adoption of the head, and a request whose T is still the head, rest on
// the signed pointer and its digest; everything else needs a promotion
diff --git a/internal/release/rollback.go b/internal/release/rollback.go
index dcdbe3831..830a0f6cf 100644
--- a/internal/release/rollback.go
+++ b/internal/release/rollback.go
@@ -66,6 +66,32 @@ func processGroup(reg *registry.Registry, bound registry.Route) []registry.Route
return out
}
+// unitServed reports whether rt's process is a unit an operator named: rt
+// names one itself, or rt is in the processGroup of a route that does. A
+// registry that cannot be listed answers true, so a RootRestartsUnits pass
+// hands the route to root rather than stamp a build the unit may not run.
+func unitServed(reg *registry.Registry, rt registry.Route) bool {
+ if rt.Unit != "" {
+ return true
+ }
+ up := upstream(rt)
+ if up == "" {
+ return false
+ }
+ all, err := reg.List()
+ if err != nil {
+ return true
+ }
+ for _, r := range all {
+ if r.Host != rt.Host && r.Unit != "" && r.App == rt.App &&
+ RouteAccount(r) == RouteAccount(rt) && upstream(r) == up &&
+ routeChannel(r) == routeChannel(rt) {
+ return true
+ }
+ }
+ return false
+}
+
// rollbackGroup puts every group route whose link names the rejected version
// onto restored (what bound now runs), stamps it to match, and returns bound
// and every group host: the hosts writeGroupRefusals refuses. restored ""
diff --git a/internal/release/rootunits_test.go b/internal/release/rootunits_test.go
new file mode 100644
index 000000000..497e81219
--- /dev/null
+++ b/internal/release/rootunits_test.go
@@ -0,0 +1,152 @@
+package release
+
+import (
+ "context"
+ "errors"
+ "testing"
+)
+
+// Issue #315: a pass that cannot restart a named unit (the agent, User=carlos)
+// hands that route to root before it stages or repoints anything. The live
+// link keeps naming the running build, the version is not poisoned for the
+// other hosts in the pass, and root's pass then installs it normally.
+func TestRootRestartsUnitsHandsNamedUnitsToRoot(t *testing.T) {
+ ctx := context.Background()
+ s := NewDirStore(t.TempDir())
+ priv, pub := testKeys(t)
+ opt := t.TempDir()
+ reg := testRegistry(t)
+ t.Cleanup(func() { reg.Close() })
+ const consoleHost, appHost = "console.example.com", "shop.example.com"
+ for _, ver := range []string{"v1", "v2"} {
+ shipRelease(t, s, "bab", "myapp", ver, adoptPlatform, []byte(ver))
+ }
+ if err := Promote(ctx, s, priv, "bab", "myapp", "canary/x", "v1"); err != nil {
+ t.Fatal(err)
+ }
+ addRoute(t, reg, consoleHost, "myapp", "canary/x", false)
+ addRoute(t, reg, appHost, "myapp", "canary/x", false)
+ if err := Pass(ctx, reg, s, pub, opt, nil); err != nil {
+ t.Fatal(err)
+ }
+ if err := reg.SetUnit(consoleHost, "console.service"); err != nil {
+ t.Fatal(err)
+ }
+ if err := Promote(ctx, s, priv, "bab", "myapp", "canary/x", "v2"); err != nil {
+ t.Fatal(err)
+ }
+
+ var restarted []string
+ restart := func(host string) error {
+ restarted = append(restarted, host)
+ if host == consoleHost {
+ return errors.New("try-restart console.service: Interactive authentication required")
+ }
+ return nil
+ }
+ type req struct {
+ host, ver string
+ err error
+ }
+ var reqs []req
+ relay := func(host, ver string, err error) { reqs = append(reqs, req{host, ver, err}) }
+
+ if err := Pass(ctx, reg, s, pub, opt, restart, OnStageBlocked(relay), RootRestartsUnits()); err != nil {
+ t.Fatal(err)
+ }
+ if len(reqs) != 1 || reqs[0].host != consoleHost || reqs[0].ver != "v2" || !errors.Is(reqs[0].err, ErrUnitRestartNeedsRoot) {
+ t.Fatalf("relays = %+v, want one for %s v2 with ErrUnitRestartNeedsRoot", reqs, consoleHost)
+ }
+ if len(restarted) != 1 || restarted[0] != appHost {
+ t.Fatalf("restarted = %v, want only %s", restarted, appHost)
+ }
+ if ver, err := LiveVersion(opt, consoleHost); err != nil || ver != "v1" {
+ t.Fatalf("%s live = %q, %v; want v1 (untouched)", consoleHost, ver, err)
+ }
+ if rt, err := reg.Get(consoleHost); err != nil || rt == nil || rt.Version != "v1" {
+ t.Fatalf("%s route = %+v, %v; want stamped v1", consoleHost, rt, err)
+ }
+ if ver, err := LiveVersion(opt, appHost); err != nil || ver != "v2" {
+ t.Fatalf("%s live = %q, %v; want v2 (not poisoned by the hand-off)", appHost, ver, err)
+ }
+ if r, err := ReadRefusal(opt, consoleHost); err != nil || r != nil {
+ t.Fatalf("refusal for %s = %+v, %v; want none", consoleHost, r, err)
+ }
+
+ // Root's pass, without the option, installs it.
+ restarted = nil
+ rootRestart := func(host string) error { restarted = append(restarted, host); return nil }
+ if err := Pass(ctx, reg, s, pub, opt, rootRestart); err != nil {
+ t.Fatal(err)
+ }
+ if ver, err := LiveVersion(opt, consoleHost); err != nil || ver != "v2" || len(restarted) != 1 || restarted[0] != consoleHost {
+ t.Fatalf("root pass: %s live = %q, %v, restarted %v; want v2 and one restart", consoleHost, ver, err, restarted)
+ }
+}
+
+// The option hands nothing over for a route this pass would not restart, and
+// leaves routes with no unit to the pass.
+func TestRootRestartsUnitsNeedsANamedUnitAndARestart(t *testing.T) {
+ ctx := context.Background()
+ s := NewDirStore(t.TempDir())
+ priv, pub := testKeys(t)
+ opt := t.TempDir()
+ reg := testRegistry(t)
+ t.Cleanup(func() { reg.Close() })
+ const host = "console.example.com"
+ shipRelease(t, s, "bab", "myapp", "v1", adoptPlatform, []byte("v1"))
+ if err := Promote(ctx, s, priv, "bab", "myapp", "canary/x", "v1"); err != nil {
+ t.Fatal(err)
+ }
+ addRoute(t, reg, host, "myapp", "canary/x", false)
+ if err := reg.SetUnit(host, "console.service"); err != nil {
+ t.Fatal(err)
+ }
+ relays := 0
+ relay := func(string, string, error) { relays++ }
+ // restart nil: a pure repoint, which needs no unit restart.
+ if err := Pass(ctx, reg, s, pub, opt, nil, OnStageBlocked(relay), RootRestartsUnits()); err != nil {
+ t.Fatal(err)
+ }
+ if ver, err := LiveVersion(opt, host); err != nil || ver != "v1" || relays != 0 {
+ t.Fatalf("live = %q, %v, relays %d; want v1 and no relay", ver, err, relays)
+ }
+}
+
+// Astra review of #315: a route with no unit of its own whose process is a
+// named unit (same account, app, channel and upstream) is root's too. Stamped
+// by the agent it would claim a build console.service does not run.
+func TestRootRestartsUnitsHandsTheProcessGroupToRoot(t *testing.T) {
+ ctx, reg, s, pub, opt := startUnitFixture(t, "console.example.com")
+ if err := reg.SetUnit("console.example.com", ""); err != nil {
+ t.Fatal(err)
+ }
+ groupFixtureRoute(t, reg, "platform.example.com", "", "127.0.0.1:8090", "console.service")
+ if err := reg.SetUpstreams("console.example.com", "127.0.0.1:8090", "", ""); err != nil {
+ t.Fatal(err)
+ }
+ if err := PointLive(opt, "platform.example.com", "bab", "old", "console"); err != nil {
+ t.Fatal(err)
+ }
+ var relayed []string
+ relay := func(host, ver string, err error) {
+ if errors.Is(err, ErrUnitRestartNeedsRoot) {
+ relayed = append(relayed, host)
+ }
+ }
+ restart := func(host string) error {
+ t.Errorf("restart(%s): every host here is the unit's", host)
+ return nil
+ }
+ if err := Pass(ctx, reg, s, pub, opt, restart, OnStageBlocked(relay), RootRestartsUnits()); err != nil {
+ t.Fatal(err)
+ }
+ if len(relayed) != 2 {
+ t.Fatalf("relayed = %v, want both hosts", relayed)
+ }
+ for _, h := range []string{"console.example.com", "platform.example.com"} {
+ if ver, err := LiveVersion(opt, h); err != nil || ver != "old" {
+ t.Fatalf("%s live = %q, %v; want old", h, ver, err)
+ }
+ }
+}