| 1 | package host |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "errors" |
| 6 | "os" |
| 7 | "path/filepath" |
| 8 | "testing" |
| 9 | |
| 10 | "github.com/carlosframework/platform/internal/appconfig" |
| 11 | "github.com/carlosframework/platform/internal/hostwrite" |
| 12 | "github.com/carlosframework/platform/internal/provision" |
| 13 | "github.com/carlosframework/platform/internal/registry" |
| 14 | "github.com/carlosframework/platform/internal/release" |
| 15 | ) |
| 16 | |
| 17 | // Harness Run 9, finding 6: a nested instance created with settings has no |
| 18 | // instance directory until its first wake, so config sync's write of its |
| 19 | // <host>.config.env failed, and only the hourly pass came back for it. The |
| 20 | // instance woke and ran without its settings (its store binding included). |
| 21 | // |
| 22 | // Through the agent's own wiring: the boot or hourly pass (syncExecSerialized) |
| 23 | // skips the row and names it, while a sibling whose directory exists gets its |
| 24 | // file in the same pass; the generation loop's tick (configGenTick) does |
| 25 | // nothing while the directory is absent, and on the first tick after the |
| 26 | // first wake has made it, writes the file. |
| 27 | func TestConfigGenTickWritesANestedRowsConfigOnceItsDirAppears(t *testing.T) { |
| 28 | configDirsPending = pendingDirs{} |
| 29 | t.Cleanup(func() { configDirsPending = pendingDirs{} }) |
| 30 | ctx := context.Background() |
| 31 | s := release.NewDirStore(t.TempDir()) |
| 32 | reg := restartTestRegistry(t) |
| 33 | dd := t.TempDir() |
| 34 | privText, _, err := appconfig.GenerateKey() |
| 35 | if err != nil { |
| 36 | t.Fatal(err) |
| 37 | } |
| 38 | priv, err := appconfig.ParsePrivateKey(privText, "test") |
| 39 | if err != nil { |
| 40 | t.Fatal(err) |
| 41 | } |
| 42 | if err := appconfig.SetEnv(ctx, s, "bab", "shop", map[string]string{"STORE_BUCKET": "store-shop"}); err != nil { |
| 43 | t.Fatal(err) |
| 44 | } |
| 45 | fresh, woken := "shop.example.com", "older.example.com" |
| 46 | for _, h := range []string{fresh, woken} { |
| 47 | if err := reg.Upsert(registry.Route{Host: h, Kind: "instance", App: "shop", Account: "bab", Hibernate: true, |
| 48 | Status: "ready", DBPath: provision.NestedDBPath(dd, h), Socket: provision.NestedSocket(h)}); err != nil { |
| 49 | t.Fatal(err) |
| 50 | } |
| 51 | } |
| 52 | // The sibling has woken before: the activator's prepare made its |
| 53 | // instance directory (os.MkdirAll(dataDir, 0o700), activator/prepare.go). |
| 54 | firstWake := func(host string) { |
| 55 | t.Helper() |
| 56 | if err := os.MkdirAll(provision.InstanceDir(dd, host), 0o700); err != nil { |
| 57 | t.Fatal(err) |
| 58 | } |
| 59 | } |
| 60 | firstWake(woken) |
| 61 | cfgPath := func(h string) string { return filepath.Join(provision.InstanceDir(dd, h), h+".config.env") } |
| 62 | |
| 63 | // The boot pass (configSyncLoop's first, a minute after start). |
| 64 | if err := syncExecSerialized(ctx, reg, s, priv, dd, nil, hostwrite.Config{}); err != nil { |
| 65 | t.Fatalf("a nested row with no directory failed the pass: %v", err) |
| 66 | } |
| 67 | if b, err := os.ReadFile(cfgPath(woken)); err != nil || string(b) != "STORE_BUCKET=store-shop\n" { |
| 68 | t.Fatalf("the sibling's bundle = %q, %v", b, err) |
| 69 | } |
| 70 | if _, err := os.Lstat(provision.InstanceDir(dd, fresh)); !os.IsNotExist(err) { |
| 71 | t.Fatalf("config sync made the instance directory itself (err %v): that is the activator's", err) |
| 72 | } |
| 73 | |
| 74 | // The generation loop: the first tick seeds its cache, and no tick |
| 75 | // writes while the directory is absent. |
| 76 | cache := newGenCache() |
| 77 | tick := func() { |
| 78 | t.Helper() |
| 79 | routes, err := reg.List() |
| 80 | if err != nil { |
| 81 | t.Fatal(err) |
| 82 | } |
| 83 | configGenTick(ctx, routes, reg, s, priv, dd, nil, hostwrite.Config{}, cache) |
| 84 | } |
| 85 | for range 2 { |
| 86 | tick() |
| 87 | } |
| 88 | if _, err := os.Lstat(cfgPath(fresh)); !os.IsNotExist(err) { |
| 89 | t.Fatalf("a bundle with nowhere to go: %v", err) |
| 90 | } |
| 91 | |
| 92 | // The first wake makes the directory; the next tick writes the bundle, |
| 93 | // with no config change and no hourly pass. |
| 94 | firstWake(fresh) |
| 95 | tick() |
| 96 | if b, err := os.ReadFile(cfgPath(fresh)); err != nil || string(b) != "STORE_BUCKET=store-shop\n" { |
| 97 | t.Fatalf("after the first wake, the bundle = %q, %v", b, err) |
| 98 | } |
| 99 | if _, ok := configDirsPending.appeared(); ok { |
| 100 | t.Fatal("the written host is still waited for") |
| 101 | } |
| 102 | // Nothing more to wait for: a later tick does not run the pass again. |
| 103 | if err := os.Remove(cfgPath(fresh)); err != nil { |
| 104 | t.Fatal(err) |
| 105 | } |
| 106 | tick() |
| 107 | if _, err := os.Lstat(cfgPath(fresh)); !os.IsNotExist(err) { |
| 108 | t.Fatalf("a tick with nothing pending ran the pass (err %v)", err) |
| 109 | } |
| 110 | } |
| 111 | |
| 112 | // A pass that fails part way does not forget a host an earlier pass was |
| 113 | // waiting for; a complete pass that no longer has the host in scope drops it. |
| 114 | func TestConfigDirsPendingKeepsWhatAFailedPassDidNotReach(t *testing.T) { |
| 115 | var p pendingDirs |
| 116 | dir := filepath.Join(t.TempDir(), "instances", "shop.example.com") |
| 117 | p.record(map[string]string{"shop.example.com": dir}, nil, true) |
| 118 | p.record(map[string]string{}, nil, false) |
| 119 | if err := os.MkdirAll(dir, 0o700); err != nil { |
| 120 | t.Fatal(err) |
| 121 | } |
| 122 | if h, ok := p.appeared(); !ok || h != "shop.example.com" { |
| 123 | t.Fatalf("appeared = %q %v after a failed pass", h, ok) |
| 124 | } |
| 125 | p.record(map[string]string{}, map[string]bool{}, true) |
| 126 | if h, ok := p.appeared(); ok { |
| 127 | t.Fatalf("appeared = %q after a complete pass with the host out of scope", h) |
| 128 | } |
| 129 | } |
| 130 | |
| 131 | // blipStore fails reads of one key while fail is set: a bucket blip on |
| 132 | // one app's bundle. |
| 133 | type blipStore struct { |
| 134 | release.Store |
| 135 | key string |
| 136 | fail bool |
| 137 | } |
| 138 | |
| 139 | func (f *blipStore) Get(ctx context.Context, key string) ([]byte, string, error) { |
| 140 | if f.fail && key == f.key { |
| 141 | return nil, "", errors.New("transient: connection reset") |
| 142 | } |
| 143 | return f.Store.Get(ctx, key) |
| 144 | } |
| 145 | |
| 146 | // Fix round 1, Astra's finding 3: the first tick after the first wake makes |
| 147 | // the directory meets a transient failure reading the app's bundle. The pass |
| 148 | // returns nil, leaving the file unwritten; the host stays waited for, and the |
| 149 | // next tick writes its config, with no config change and no hourly pass. A |
| 150 | // bundle that keeps failing is retried on a backoff, not every tick. |
| 151 | func TestConfigGenTickRetriesAHostAPassLeftUndone(t *testing.T) { |
| 152 | configDirsPending = pendingDirs{} |
| 153 | t.Cleanup(func() { configDirsPending = pendingDirs{} }) |
| 154 | ctx := context.Background() |
| 155 | base := release.NewDirStore(t.TempDir()) |
| 156 | s := &blipStore{Store: base, key: appconfig.Key("bab", "shop")} |
| 157 | reg := restartTestRegistry(t) |
| 158 | dd := t.TempDir() |
| 159 | privText, _, err := appconfig.GenerateKey() |
| 160 | if err != nil { |
| 161 | t.Fatal(err) |
| 162 | } |
| 163 | priv, err := appconfig.ParsePrivateKey(privText, "test") |
| 164 | if err != nil { |
| 165 | t.Fatal(err) |
| 166 | } |
| 167 | if err := appconfig.SetEnv(ctx, base, "bab", "shop", map[string]string{"STORE_BUCKET": "store-shop"}); err != nil { |
| 168 | t.Fatal(err) |
| 169 | } |
| 170 | host := "shop.example.com" |
| 171 | if err := reg.Upsert(registry.Route{Host: host, Kind: "instance", App: "shop", Account: "bab", Hibernate: true, |
| 172 | Status: "ready", DBPath: provision.NestedDBPath(dd, host), Socket: provision.NestedSocket(host)}); err != nil { |
| 173 | t.Fatal(err) |
| 174 | } |
| 175 | cfgPath := filepath.Join(provision.InstanceDir(dd, host), host+".config.env") |
| 176 | cache := newGenCache() |
| 177 | passes := 0 |
| 178 | tick := func() { |
| 179 | t.Helper() |
| 180 | routes, err := reg.List() |
| 181 | if err != nil { |
| 182 | t.Fatal(err) |
| 183 | } |
| 184 | if _, due := configDirsPending.peek(); due { |
| 185 | passes++ |
| 186 | } |
| 187 | configGenTick(ctx, routes, reg, s, priv, dd, nil, hostwrite.Config{}, cache) |
| 188 | } |
| 189 | if err := syncExecSerialized(ctx, reg, s, priv, dd, nil, hostwrite.Config{}); err != nil { |
| 190 | t.Fatal(err) |
| 191 | } |
| 192 | tick() // seeds the cache |
| 193 | if err := os.MkdirAll(provision.InstanceDir(dd, host), 0o700); err != nil { |
| 194 | t.Fatal(err) |
| 195 | } |
| 196 | s.fail = true |
| 197 | tick() |
| 198 | if _, err := os.Lstat(cfgPath); !os.IsNotExist(err) { |
| 199 | t.Fatalf("a bundle that did not read was written (err %v)", err) |
| 200 | } |
| 201 | if _, ok := configDirsPending.peek(); !ok { |
| 202 | t.Fatal("the pass that left the host undone dropped it") |
| 203 | } |
| 204 | s.fail = false |
| 205 | tick() |
| 206 | if b, err := os.ReadFile(cfgPath); err != nil || string(b) != "STORE_BUCKET=store-shop\n" { |
| 207 | t.Fatalf("the next tick's bundle = %q, %v", b, err) |
| 208 | } |
| 209 | if _, ok := configDirsPending.peek(); ok { |
| 210 | t.Fatal("the written host is still waited for") |
| 211 | } |
| 212 | |
| 213 | // A bundle that keeps failing: retries back off (1, 2, 4 ... ticks apart). |
| 214 | if err := os.Remove(cfgPath); err != nil { |
| 215 | t.Fatal(err) |
| 216 | } |
| 217 | configDirsPending.record(map[string]string{host: provision.InstanceDir(dd, host)}, nil, false) |
| 218 | s.fail = true |
| 219 | passes = 0 |
| 220 | for range 20 { |
| 221 | tick() |
| 222 | } |
| 223 | if passes == 0 || passes > 6 { |
| 224 | t.Fatalf("%d retries in 20 ticks of a failing bundle, want a backoff (1 to 6)", passes) |
| 225 | } |
| 226 | } |
| 227 | |
| 228 | // peek is appeared without the tick: whether a retry is due now. |
| 229 | func (p *pendingDirs) peek() (string, bool) { |
| 230 | p.mu.Lock() |
| 231 | defer p.mu.Unlock() |
| 232 | for h, e := range p.dirs { |
| 233 | if _, err := os.Lstat(e.dir); err == nil && e.waitTicks == 0 { |
| 234 | return h, true |
| 235 | } |
| 236 | } |
| 237 | return "", false |
| 238 | } |
| 239 | |
| 240 | // TestPendingDirsBackoffSaturates: a host that keeps failing backs off to |
| 241 | // the cap and stays there; the failure count never grows far enough for the |
| 242 | // shift to overflow into a negative or zero delay (Astra fix round 1). |
| 243 | func TestPendingDirsBackoffSaturates(t *testing.T) { |
| 244 | var p pendingDirs |
| 245 | dir := t.TempDir() |
| 246 | p.record(map[string]string{"shop.example.com": dir}, nil, false) |
| 247 | for i := 0; i < 200; i++ { |
| 248 | p.record(nil, map[string]bool{"shop.example.com": false}, false) |
| 249 | e := p.dirs["shop.example.com"] |
| 250 | if e == nil { |
| 251 | t.Fatalf("pass %d: the waiting host was dropped", i) |
| 252 | } |
| 253 | if e.waitTicks < 0 || e.waitTicks > pendingRetryCap-1 { |
| 254 | t.Fatalf("pass %d: waitTicks %d outside [0,%d]", i, e.waitTicks, pendingRetryCap-1) |
| 255 | } |
| 256 | } |
| 257 | if got := p.dirs["shop.example.com"].waitTicks; got != pendingRetryCap-1 { |
| 258 | t.Fatalf("after 200 failures waitTicks = %d, want the cap %d", got, pendingRetryCap-1) |
| 259 | } |
| 260 | p.record(nil, map[string]bool{"shop.example.com": true}, false) |
| 261 | if _, ok := p.dirs["shop.example.com"]; ok { |
| 262 | t.Fatal("a delivered host stays pending") |
| 263 | } |
| 264 | } |
| 265 | |