detect upstream drift and arch-any moves via live pacman db

2026-05-19 02:01:34 +02:00
parent a8085c9d19
commit 6b524f30db
4 changed files with 121 additions and 5 deletions
@@ -4,6 +4,7 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"github.com/Jguer/go-alpm/v2"
 	"github.com/c2h5oh/datasize"
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/sethvargo/go-retry"
@@ -464,14 +465,52 @@ func (b *BuildManager) genQueue(ctx context.Context) ([]*ProtoPackage, error) {
 				}
 			}

-			if pkg.DBPackage.TagRev != nil && *pkg.DBPackage.TagRev == state.TagRev {
+			// Cross-check against the live pacman DB: state.git lags
+			// reality for pkgrel-only rebuilds and arch-any moves leave
+			// stale x86_64 state files behind. SyncPkg was resolved by
+			// isAvailable above; reuse it.
+			if pkg.SyncPkg != nil {
+				if pkg.SyncPkg.Architecture() == "any" && pkg.Arch == "x86_64" {
+					// Already marked from a prior cycle: nothing to redo.
+					if pkg.DBPackage.Status == dbpackage.StatusSkipped &&
+						pkg.DBPackage.SkipReason == SkipReasonAnyArchMoved {
+						continue
+					}
+					log.Infof("[QG] %s->%s upstream moved to any-arch, dropping", pkg.FullRepo, pkg.Pkgbase)
+					pkg.DBPackage, err = pkg.DBPackage.Update().
+						SetStatus(dbpackage.StatusSkipped).
+						SetSkipReason(SkipReasonAnyArchMoved).
+						SetTagRev(state.TagRev).
+						SetVersion(state.PkgVer).
+						Save(ctx)
+					if err != nil {
+						log.Warningf("[QG] error updating dbpackage %s: %v", state.Pkgbase, err)
+						continue
+					}
+					buildManager.repoPurge[pkg.FullRepo] <- []*ProtoPackage{pkg}
+					continue
+				}
+				if alpm.VerCmp(pkg.SyncPkg.Version(), state.PkgVer) > 0 {
+					log.Infof("[QG] %s->%s upstream version drift detected (state: %s < upstream: %s), building from %s",
+						pkg.FullRepo, pkg.Pkgbase, state.PkgVer, pkg.SyncPkg.Version(), upstreamDefaultGitBranch)
+					pkg.UseLatest = true
+				}
+			}
+
+			if !pkg.UseLatest && pkg.DBPackage.TagRev != nil && *pkg.DBPackage.TagRev == state.TagRev {
 				continue
 			}

-			// try download .SRCINFO from repo
-			srcInfo, err := downloadSRCINFO(pkg.DBPackage.Pkgbase, state.TagRev)
+			srcRef := state.TagRev
+			if pkg.UseLatest {
+				srcRef = upstreamDefaultGitBranch
+			}
+			srcInfo, err := downloadSRCINFO(pkg.DBPackage.Pkgbase, srcRef)
 			if err == nil {
 				pkg.Srcinfo = srcInfo
+				if pkg.UseLatest {
+					pkg.Version = constructVersion(srcInfo.Pkgver, srcInfo.Pkgrel, srcInfo.Epoch)
+				}
 			}

 			if !pkg.isEligible(ctx) {
@@ -2,6 +2,7 @@ package main

 import (
 	"context"
+	"github.com/Jguer/go-alpm/v2"
 	log "github.com/sirupsen/logrus"
 	"os"
 	"os/exec"
@@ -146,6 +147,26 @@ func housekeeping(ctx context.Context, repo, march string, wg *sync.WaitGroup) e
 				return err
 			}
 		}
+
+		// detect upstream version drift: Arch sometimes ships a pkgrel
+		// rebuild without updating state.git, so state.TagRev never
+		// changes and genQueue never re-queues. Compare directly against
+		// the pacman sync DB (always current) and force a re-queue.
+		// Only act on settled packages; a Queued/Building/Delayed row is
+		// already moving and our SetStatus(Queued) would race it.
+		if pkg.DBPackage.Status == dbpackage.StatusLatest &&
+			pkg.DBPackage.Version != "" && pkgResolved.Version() != "" &&
+			alpm.VerCmp(pkgResolved.Version(), pkg.DBPackage.Version) > 0 {
+			log.Infof("[HK] %s->%s upstream version drift detected (db: %s < upstream: %s), requeuing",
+				pkg.FullRepo, pkg.Pkgbase, pkg.DBPackage.Version, pkgResolved.Version())
+			pkg.DBPackage, err = pkg.DBPackage.Update().
+				SetStatus(dbpackage.StatusQueued).
+				ClearTagRev().
+				Save(ctx)
+			if err != nil {
+				return err
+			}
+		}
 	}

 	// check all packages from db for existence
@@ -24,6 +24,14 @@ import (
 	"time"
 )

+// Skip reasons surfaced from the build pipeline. Hoisted so the same string
+// is used wherever a package is skipped for that reason.
+const (
+	SkipReasonAnyArch        = "arch = any"
+	SkipReasonAnyArchMoved   = "arch = any (moved upstream)"
+	upstreamDefaultGitBranch = "main"
+)
+
 type ProtoPackage struct {
 	Pkgbase   string
 	Srcinfo   *srcinfo.Srcinfo
@@ -36,6 +44,13 @@ type ProtoPackage struct {
 	DBPackage *ent.DBPackage
 	Pkgbuild  string
 	State     *StateInfo
+	// SyncPkg is the alpm package resolved by isAvailable; cached so
+	// later checks (drift, arch-any-move) don't re-do FindSatisfier.
+	SyncPkg alpm.IPackage
+	// UseLatest causes SRCINFO/PKGBUILD to be fetched from
+	// upstreamDefaultGitBranch (main) instead of state.TagVer/state.TagRev.
+	// Set when upstream Arch ships a rebuild that state.git did not record.
+	UseLatest bool
 }

 var (
@@ -47,7 +62,7 @@ func (p *ProtoPackage) isEligible(ctx context.Context) bool {
 	switch {
 	case p.Arch == "any":
 		log.Debugf("skipped %s: any-package", p.Pkgbase)
-		p.DBPackage.SkipReason = "arch = any"
+		p.DBPackage.SkipReason = SkipReasonAnyArch
 		p.DBPackage.Status = dbpackage.StatusSkipped
 		skipping = true
 	case MatchGlobList(p.Pkgbase, conf.Blacklist.Packages):
@@ -365,7 +380,7 @@ func (p *ProtoPackage) setupBuildDir(ctx context.Context) (string, error) {
 	gr = retry.WithMaxRetries(conf.MaxCloneRetries, gr)

 	if err := retry.Do(ctx, gr, func(ctx context.Context) error {
-		cmd := exec.CommandContext(ctx, "git", "clone", "--depth", "1", "--branch", p.State.TagVer, //nolint:gosec
+		cmd := exec.CommandContext(ctx, "git", "clone", "--depth", "1", "--branch", p.cloneBranch(), //nolint:gosec
 			fmt.Sprintf("https://gitlab.archlinux.org/archlinux/packaging/packages/%s.git", gitlabPath), buildDir)
 		res, err := cmd.CombinedOutput()
 		log.Debug(string(res))
@@ -541,9 +556,21 @@ func (p *ProtoPackage) isAvailable(ctx context.Context, h *alpm.Handle) bool {
 		return false
 	}

+	p.SyncPkg = pkg
 	return true
 }

+// cloneBranch returns the git ref to use when cloning the upstream
+// packaging repo. The default ref is the tag recorded in state.git;
+// UseLatest overrides that to track the package repo's main branch
+// (used when state.git lags real upstream).
+func (p *ProtoPackage) cloneBranch() string {
+	if p.UseLatest {
+		return upstreamDefaultGitBranch
+	}
+	return p.State.TagVer
+}
+
 func (p *ProtoPackage) GitVersion(h *alpm.Handle) (string, error) {
 	if p.Pkgbase == "" {
 		return "", errors.New("invalid arguments")
@@ -96,6 +96,35 @@ package() {
 # vim:set sw=2 et:
 `

+func TestCloneBranch(t *testing.T) { //nolint:paralleltest
+	for _, tc := range []struct {
+		name      string
+		useLatest bool
+		tagVer    string
+		want      string
+	}{
+		{"state-tag-default", false, "3.3-4", "3.3-4"},
+		{"main-when-drift", true, "3.3-4", "main"},
+		{"main-ignores-empty-tag", true, "", "main"},
+	} {
+		t.Run(tc.name, func(t *testing.T) {
+			p := &ProtoPackage{
+				UseLatest: tc.useLatest,
+				State:     &StateInfo{TagVer: tc.tagVer},
+			}
+			if got := p.cloneBranch(); got != tc.want {
+				t.Errorf("cloneBranch() = %q, want %q", got, tc.want)
+			}
+		})
+	}
+}
+
+func TestSkipReasonsDistinct(t *testing.T) { //nolint:paralleltest
+	if SkipReasonAnyArch == SkipReasonAnyArchMoved {
+		t.Fatal("any-arch skip reasons must remain distinguishable: callers and operators inspect skip_reason to tell why a package was dropped")
+	}
+}
+
 func TestIncreasePkgRel(t *testing.T) { //nolint:paralleltest
 	pkgbuild, err := os.CreateTemp(t.TempDir(), "")
 	if err != nil {