| 1 | package host |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "encoding/json" |
| 6 | "errors" |
| 7 | "os" |
| 8 | "strings" |
| 9 | |
| 10 | "github.com/carlosframework/platform/internal/hostwrite" |
| 11 | "github.com/carlosframework/platform/internal/release" |
| 12 | "github.com/carlosframework/platform/internal/rootstate" |
| 13 | ) |
| 14 | |
| 15 | func recoverRootRestarts(ctx context.Context, sys unitSystem, optDir string, getenv func(string) string, checks rootRestartChecks, writes hostwrite.Config) error { |
| 16 | legacy, err := checks.allowLegacy() |
| 17 | if err != nil { |
| 18 | return err |
| 19 | } |
| 20 | if legacy { |
| 21 | checks.noteLegacy() |
| 22 | // D3 still records root health without a census. Preparing its |
| 23 | // public parent must not create or recover bound restart state. |
| 24 | return rootstate.EnsurePublicDir(release.RootHealthyDir) |
| 25 | } |
| 26 | if checks.attempts == nil || checks.prepareRollback == nil { |
| 27 | return os.ErrPermission |
| 28 | } |
| 29 | s, err := checks.attempts() |
| 30 | if err != nil { |
| 31 | return err |
| 32 | } |
| 33 | defer s.Close() |
| 34 | names, err := s.Names() |
| 35 | if err != nil { |
| 36 | return err |
| 37 | } |
| 38 | var failures []error |
| 39 | for _, name := range names { |
| 40 | if !strings.HasSuffix(name, ".attempt") { |
| 41 | continue |
| 42 | } |
| 43 | body, err := s.Read(name) |
| 44 | var a rollbackAttempt |
| 45 | if err != nil || json.Unmarshal(body, &a) != nil || name != rollbackJournalKey(a.Host) { |
| 46 | failures = append(failures, errors.Join(err, os.ErrPermission)) |
| 47 | continue |
| 48 | } |
| 49 | if a.Settled && a.History == nil { |
| 50 | if err := removeRollbackCapabilities(s, a.Host); err != nil { |
| 51 | failures = append(failures, err) |
| 52 | } |
| 53 | continue |
| 54 | } |
| 55 | if err := resumeRootRestart(ctx, sys, optDir, getenv, checks, writes, s, a); err != nil { |
| 56 | failures = append(failures, err) |
| 57 | } |
| 58 | } |
| 59 | return errors.Join(failures...) |
| 60 | } |
| 61 | |
| 62 | func resumeRootRestart(ctx context.Context, sys unitSystem, optDir string, getenv func(string) string, checks rootRestartChecks, writes hostwrite.Config, s rootstate.Store, expected rollbackAttempt) (result error) { |
| 63 | unlock, err := s.Lock(ctx, rollbackHostKey(expected.Host)+".lock") |
| 64 | if err != nil { |
| 65 | return err |
| 66 | } |
| 67 | defer unlock() |
| 68 | current, err := readRollbackAttempt(s, expected.Host) |
| 69 | if err != nil || current.ID != expected.ID { |
| 70 | return errors.Join(err, os.ErrPermission) |
| 71 | } |
| 72 | if current.History != nil { |
| 73 | return resumeHistoryRestart(ctx, sys, optDir, getenv, checks, writes, s, current) |
| 74 | } |
| 75 | if current.Settled { |
| 76 | return nil |
| 77 | } |
| 78 | a, b, err := loadRollback(s, expected.Host, checks.now()) |
| 79 | if a.ID != expected.ID { |
| 80 | return errors.Join(err, os.ErrPermission) |
| 81 | } |
| 82 | defer func() { result = errors.Join(result, settleRollbackAttempt(s, a)) }() |
| 83 | if err != nil { |
| 84 | return err |
| 85 | } |
| 86 | w, err := writes.Select(ctx, a.Host, "release") |
| 87 | if err != nil { |
| 88 | return err |
| 89 | } |
| 90 | checks.identity = a.Identity |
| 91 | checks.effect = func(fn func() error) error { return w.With(ctx, a.Identity, fn) } |
| 92 | err = checks.effect(func() error { |
| 93 | var err error |
| 94 | b, err = checks.prepareRollback(ctx, b) |
| 95 | if err != nil { |
| 96 | return err |
| 97 | } |
| 98 | p, err := sys.Show(b.Route.Unit) |
| 99 | if err != nil { |
| 100 | return err |
| 101 | } |
| 102 | if p.ActiveState == "inactive" { |
| 103 | return errUnitStopped |
| 104 | } |
| 105 | live, err := checks.live(optDir, a.Host) |
| 106 | old := release.VersionPath(optDir, b.Route.Account, b.Version, b.Route.App) |
| 107 | if err != nil || (live != a.Target && live != old) { |
| 108 | return os.ErrPermission |
| 109 | } |
| 110 | if !a.Recovering && !a.Failed && p.ActiveState == "active" && p.MainPID > 0 && live == a.Target { |
| 111 | checks.verifyOnly = true |
| 112 | checks.targetDigest = a.TargetDigest |
| 113 | return nil |
| 114 | } |
| 115 | if a.Recovering && p.ActiveState == "active" && p.MainPID > 0 && live == old { |
| 116 | checks.verifyOnly = true |
| 117 | } else if p.ActiveState != "failed" && p.MainPID != 0 && !a.Failed { |
| 118 | return os.ErrPermission |
| 119 | } |
| 120 | a.Recovering = true |
| 121 | if err := writeRollbackState(s, rollbackJournalKey(a.Host), a); err != nil { |
| 122 | return err |
| 123 | } |
| 124 | if live != old { |
| 125 | return release.PointLive(optDir, a.Host, b.Route.Account, b.Version, b.Route.App) |
| 126 | } |
| 127 | return nil |
| 128 | }) |
| 129 | if err != nil { |
| 130 | return err |
| 131 | } |
| 132 | if checks.verifyOnly && !a.Recovering { |
| 133 | b, err = checks.host(ctx, a.Host) |
| 134 | if err != nil { |
| 135 | return err |
| 136 | } |
| 137 | return restartBoundUnit(sys, optDir, getenv, b, false, nil, checks) |
| 138 | } |
| 139 | b, err = checks.rollback(ctx, b) |
| 140 | if err != nil { |
| 141 | return err |
| 142 | } |
| 143 | return restartBoundUnit(sys, optDir, getenv, b, true, nil, checks) |
| 144 | } |
| 145 | |