| 1 | //go:build cgo |
| 2 | |
| 3 | package as400 |
| 4 | |
| 5 | import ( |
| 6 | "context" |
| 7 | "errors" |
| 8 | "fmt" |
| 9 | "maps" |
| 10 | "strings" |
| 11 | "sync" |
| 12 | "time" |
| 13 | |
| 14 | "golang.org/x/sync/errgroup" |
| 15 | |
| 16 | as400proto "github.com/netdata/netdata/go/plugins/plugin/ibm.d/protocols/as400" |
| 17 | ) |
| 18 | |
| 19 | type slowPathConfig struct { |
| 20 | enabled bool |
| 21 | interval time.Duration |
| 22 | maxConnections int |
| 23 | } |
| 24 | |
| 25 | type messageQueueSnapshot struct { |
| 26 | metrics map[string]messageQueueInstanceMetrics |
| 27 | meta map[string]messageQueueMetrics |
| 28 | timestamp time.Time |
| 29 | err error |
| 30 | } |
| 31 | |
| 32 | type jobQueueSnapshot struct { |
| 33 | metrics map[string]jobQueueInstanceMetrics |
| 34 | meta map[string]jobQueueMetrics |
| 35 | timestamp time.Time |
| 36 | err error |
| 37 | } |
| 38 | |
| 39 | type outputQueueSnapshot struct { |
| 40 | metrics map[string]outputQueueInstanceMetrics |
| 41 | meta map[string]outputQueueMetrics |
| 42 | timestamp time.Time |
| 43 | err error |
| 44 | } |
| 45 | |
| 46 | type subsystemSnapshot struct { |
| 47 | metrics map[string]subsystemInstanceMetrics |
| 48 | meta map[string]subsystemMetrics |
| 49 | timestamp time.Time |
| 50 | err error |
| 51 | } |
| 52 | |
| 53 | type planCacheSnapshot struct { |
| 54 | values map[string]planCacheInstanceMetrics |
| 55 | meta map[string]planCacheMetrics |
| 56 | timestamp time.Time |
| 57 | err error |
| 58 | } |
| 59 | |
| 60 | type slowCache struct { |
| 61 | mu sync.RWMutex |
| 62 | messageQueues messageQueueSnapshot |
| 63 | jobQueues jobQueueSnapshot |
| 64 | outputQueues outputQueueSnapshot |
| 65 | subsystems subsystemSnapshot |
| 66 | planCache planCacheSnapshot |
| 67 | latency latencyCache |
| 68 | } |
| 69 | |
| 70 | func (c *Collector) slowPathActive() bool { |
| 71 | return c != nil && c.slow.config.enabled && c.slow.client != nil |
| 72 | } |
| 73 | |
| 74 | func (c *slowCache) beginLatencyCycle(ts time.Time) { |
| 75 | c.latency.beginCycle(ts) |
| 76 | } |
| 77 | |
| 78 | func (c *slowCache) addLatency(name string, value int64) { |
| 79 | c.latency.add(name, value) |
| 80 | } |
| 81 | |
| 82 | func (c *slowCache) setMessageQueues(snapshot messageQueueSnapshot) { |
| 83 | c.mu.Lock() |
| 84 | c.messageQueues = snapshot |
| 85 | c.mu.Unlock() |
| 86 | } |
| 87 | |
| 88 | func (c *slowCache) setJobQueues(snapshot jobQueueSnapshot) { |
| 89 | c.mu.Lock() |
| 90 | c.jobQueues = snapshot |
| 91 | c.mu.Unlock() |
| 92 | } |
| 93 | |
| 94 | func (c *slowCache) setOutputQueues(snapshot outputQueueSnapshot) { |
| 95 | c.mu.Lock() |
| 96 | c.outputQueues = snapshot |
| 97 | c.mu.Unlock() |
| 98 | } |
| 99 | |
| 100 | func (c *slowCache) setSubsystems(snapshot subsystemSnapshot) { |
| 101 | c.mu.Lock() |
| 102 | c.subsystems = snapshot |
| 103 | c.mu.Unlock() |
| 104 | } |
| 105 | |
| 106 | func (c *slowCache) setPlanCache(snapshot planCacheSnapshot) { |
| 107 | c.mu.Lock() |
| 108 | c.planCache = snapshot |
| 109 | c.mu.Unlock() |
| 110 | } |
| 111 | |
| 112 | func (c *slowCache) getMessageQueues() messageQueueSnapshot { |
| 113 | c.mu.RLock() |
| 114 | defer c.mu.RUnlock() |
| 115 | return cloneMessageQueueSnapshot(c.messageQueues) |
| 116 | } |
| 117 | |
| 118 | func (c *slowCache) getJobQueues() jobQueueSnapshot { |
| 119 | c.mu.RLock() |
| 120 | defer c.mu.RUnlock() |
| 121 | return cloneJobQueueSnapshot(c.jobQueues) |
| 122 | } |
| 123 | |
| 124 | func (c *slowCache) getOutputQueues() outputQueueSnapshot { |
| 125 | c.mu.RLock() |
| 126 | defer c.mu.RUnlock() |
| 127 | return cloneOutputQueueSnapshot(c.outputQueues) |
| 128 | } |
| 129 | |
| 130 | func (c *slowCache) getSubsystems() subsystemSnapshot { |
| 131 | c.mu.RLock() |
| 132 | defer c.mu.RUnlock() |
| 133 | return cloneSubsystemSnapshot(c.subsystems) |
| 134 | } |
| 135 | |
| 136 | func (c *slowCache) getPlanCache() planCacheSnapshot { |
| 137 | c.mu.RLock() |
| 138 | defer c.mu.RUnlock() |
| 139 | return clonePlanCacheSnapshot(c.planCache) |
| 140 | } |
| 141 | |
| 142 | func (c *slowCache) getLatencies() (map[string]int64, time.Time) { |
| 143 | return c.latency.snapshot() |
| 144 | } |
| 145 | |
| 146 | func cloneMessageQueueSnapshot(src messageQueueSnapshot) messageQueueSnapshot { |
| 147 | dst := messageQueueSnapshot{ |
| 148 | timestamp: src.timestamp, |
| 149 | err: src.err, |
| 150 | } |
| 151 | if src.metrics != nil { |
| 152 | dst.metrics = make(map[string]messageQueueInstanceMetrics, len(src.metrics)) |
| 153 | maps.Copy(dst.metrics, src.metrics) |
| 154 | } |
| 155 | if src.meta != nil { |
| 156 | dst.meta = make(map[string]messageQueueMetrics, len(src.meta)) |
| 157 | maps.Copy(dst.meta, src.meta) |
| 158 | } |
| 159 | return dst |
| 160 | } |
| 161 | |
| 162 | func cloneJobQueueSnapshot(src jobQueueSnapshot) jobQueueSnapshot { |
| 163 | dst := jobQueueSnapshot{ |
| 164 | timestamp: src.timestamp, |
| 165 | err: src.err, |
| 166 | } |
| 167 | if src.metrics != nil { |
| 168 | dst.metrics = make(map[string]jobQueueInstanceMetrics, len(src.metrics)) |
| 169 | maps.Copy(dst.metrics, src.metrics) |
| 170 | } |
| 171 | if src.meta != nil { |
| 172 | dst.meta = make(map[string]jobQueueMetrics, len(src.meta)) |
| 173 | maps.Copy(dst.meta, src.meta) |
| 174 | } |
| 175 | return dst |
| 176 | } |
| 177 | |
| 178 | func cloneOutputQueueSnapshot(src outputQueueSnapshot) outputQueueSnapshot { |
| 179 | dst := outputQueueSnapshot{ |
| 180 | timestamp: src.timestamp, |
| 181 | err: src.err, |
| 182 | } |
| 183 | if src.metrics != nil { |
| 184 | dst.metrics = make(map[string]outputQueueInstanceMetrics, len(src.metrics)) |
| 185 | maps.Copy(dst.metrics, src.metrics) |
| 186 | } |
| 187 | if src.meta != nil { |
| 188 | dst.meta = make(map[string]outputQueueMetrics, len(src.meta)) |
| 189 | maps.Copy(dst.meta, src.meta) |
| 190 | } |
| 191 | return dst |
| 192 | } |
| 193 | |
| 194 | func cloneSubsystemSnapshot(src subsystemSnapshot) subsystemSnapshot { |
| 195 | dst := subsystemSnapshot{ |
| 196 | timestamp: src.timestamp, |
| 197 | err: src.err, |
| 198 | } |
| 199 | if src.metrics != nil { |
| 200 | dst.metrics = make(map[string]subsystemInstanceMetrics, len(src.metrics)) |
| 201 | maps.Copy(dst.metrics, src.metrics) |
| 202 | } |
| 203 | if src.meta != nil { |
| 204 | dst.meta = make(map[string]subsystemMetrics, len(src.meta)) |
| 205 | maps.Copy(dst.meta, src.meta) |
| 206 | } |
| 207 | return dst |
| 208 | } |
| 209 | |
| 210 | func clonePlanCacheSnapshot(src planCacheSnapshot) planCacheSnapshot { |
| 211 | dst := planCacheSnapshot{ |
| 212 | timestamp: src.timestamp, |
| 213 | err: src.err, |
| 214 | } |
| 215 | if src.values != nil { |
| 216 | dst.values = make(map[string]planCacheInstanceMetrics, len(src.values)) |
| 217 | maps.Copy(dst.values, src.values) |
| 218 | } |
| 219 | if src.meta != nil { |
| 220 | dst.meta = make(map[string]planCacheMetrics, len(src.meta)) |
| 221 | maps.Copy(dst.meta, src.meta) |
| 222 | } |
| 223 | return dst |
| 224 | } |
| 225 | |
| 226 | func (c *Collector) startSlowPath() error { |
| 227 | c.stopSlowPath() |
| 228 | |
| 229 | cfg := slowPathConfig{ |
| 230 | enabled: c.SlowPath, |
| 231 | interval: time.Duration(c.SlowPathUpdateEvery), |
| 232 | maxConnections: c.SlowPathMaxConnections, |
| 233 | } |
| 234 | |
| 235 | if !cfg.enabled { |
| 236 | c.Debugf("slow path disabled; running sequential-only mode") |
| 237 | c.slow.config = cfg |
| 238 | return nil |
| 239 | } |
| 240 | |
| 241 | if cfg.interval <= 0 { |
| 242 | cfg.interval = 30 * time.Second |
| 243 | } |
| 244 | if cfg.maxConnections <= 0 { |
| 245 | cfg.maxConnections = 1 |
| 246 | } |
| 247 | |
| 248 | fastInterval := time.Duration(c.fastPathIntervalSeconds()) * time.Second |
| 249 | if fastInterval <= 0 { |
| 250 | fastInterval = time.Second |
| 251 | } |
| 252 | if cfg.interval < fastInterval { |
| 253 | c.Warningf("slow path update every %s is shorter than main update %s; using %s", cfg.interval, fastInterval, fastInterval) |
| 254 | cfg.interval = fastInterval |
| 255 | } |
| 256 | |
| 257 | clientCfg := as400proto.Config{ |
| 258 | DSN: c.DSN, |
| 259 | Timeout: time.Duration(c.Timeout), |
| 260 | MaxOpenConns: cfg.maxConnections, |
| 261 | } |
| 262 | |
| 263 | client := as400proto.NewClient(clientCfg) |
| 264 | ctx := context.Background() |
| 265 | if err := client.Connect(ctx); err != nil { |
| 266 | return fmt.Errorf("slow path: connect failed: %w", err) |
| 267 | } |
| 268 | if err := client.Ping(ctx); err != nil { |
| 269 | _ = client.Close() |
| 270 | return fmt.Errorf("slow path: ping failed: %w", err) |
| 271 | } |
| 272 | |
| 273 | runCtx, cancel := context.WithCancel(context.Background()) |
| 274 | c.slow.client = client |
| 275 | c.slow.cancel = cancel |
| 276 | c.slow.config = cfg |
| 277 | c.slow.wg.Add(1) |
| 278 | go c.runSlowPath(runCtx) |
| 279 | c.Infof("slow path worker started (interval=%s, max_conns=%d)", cfg.interval, cfg.maxConnections) |
| 280 | return nil |
| 281 | } |
| 282 | |
| 283 | func (c *Collector) stopSlowPath() { |
| 284 | if c.slow.cancel != nil { |
| 285 | c.slow.cancel() |
| 286 | } |
| 287 | c.slow.wg.Wait() |
| 288 | if c.slow.client != nil { |
| 289 | if err := c.slow.client.Close(); err != nil { |
| 290 | c.Errorf("slow path: closing client failed: %v", err) |
| 291 | } |
| 292 | } |
| 293 | c.slow.cancel = nil |
| 294 | c.slow.client = nil |
| 295 | c.slow.config = slowPathConfig{} |
| 296 | c.slow.cache = slowCache{} |
| 297 | } |
| 298 | |
| 299 | func (c *Collector) runSlowPath(ctx context.Context) { |
| 300 | defer c.slow.wg.Done() |
| 301 | |
| 302 | interval := c.slow.config.interval |
| 303 | if interval <= 0 { |
| 304 | interval = 30 * time.Second |
| 305 | } |
| 306 | |
| 307 | now := time.Now() |
| 308 | beat := now |
| 309 | c.runSlowCollectors(ctx, beat) |
| 310 | nextBeat := beat.Add(interval) |
| 311 | |
| 312 | for { |
| 313 | sleep := time.Until(nextBeat) |
| 314 | if sleep > 0 { |
| 315 | timer := time.NewTimer(sleep) |
| 316 | select { |
| 317 | case <-ctx.Done(): |
| 318 | timer.Stop() |
| 319 | return |
| 320 | case <-timer.C: |
| 321 | } |
| 322 | } else { |
| 323 | select { |
| 324 | case <-ctx.Done(): |
| 325 | return |
| 326 | default: |
| 327 | } |
| 328 | } |
| 329 | |
| 330 | beat = nextBeat |
| 331 | c.runSlowCollectors(ctx, beat) |
| 332 | |
| 333 | nextBeat = nextBeat.Add(interval) |
| 334 | now = time.Now() |
| 335 | for nextBeat.Before(now) { |
| 336 | nextBeat = nextBeat.Add(interval) |
| 337 | } |
| 338 | } |
| 339 | } |
| 340 | |
| 341 | func (c *Collector) runSlowCollectors(ctx context.Context, beat time.Time) { |
| 342 | if ctx.Err() != nil { |
| 343 | return |
| 344 | } |
| 345 | |
| 346 | c.slow.cache.beginLatencyCycle(beat) |
| 347 | |
| 348 | workCtx, cancel := context.WithCancel(ctx) |
| 349 | defer cancel() |
| 350 | |
| 351 | group, groupCtx := errgroup.WithContext(workCtx) |
| 352 | group.SetLimit(c.slow.config.maxConnections) |
| 353 | |
| 354 | group.Go(func() error { |
| 355 | snapshot, err := c.fetchMessageQueues(groupCtx, beat, c.slowDoQuery) |
| 356 | c.slow.cache.setMessageQueues(snapshot) |
| 357 | if err != nil { |
| 358 | return fmt.Errorf("message queues: %w", err) |
| 359 | } |
| 360 | return nil |
| 361 | }) |
| 362 | |
| 363 | group.Go(func() error { |
| 364 | snapshot, err := c.fetchJobQueues(groupCtx, beat, c.slowDoQuery) |
| 365 | c.slow.cache.setJobQueues(snapshot) |
| 366 | if err != nil { |
| 367 | return fmt.Errorf("job queues: %w", err) |
| 368 | } |
| 369 | return nil |
| 370 | }) |
| 371 | |
| 372 | group.Go(func() error { |
| 373 | snapshot, err := c.fetchOutputQueues(groupCtx, beat, c.slowDoQuery) |
| 374 | c.slow.cache.setOutputQueues(snapshot) |
| 375 | if err != nil { |
| 376 | return fmt.Errorf("output queues: %w", err) |
| 377 | } |
| 378 | return nil |
| 379 | }) |
| 380 | |
| 381 | group.Go(func() error { |
| 382 | snapshot, err := c.fetchSubsystems(groupCtx, beat, c.slowDoQuery, c.slowDoQueryRow) |
| 383 | c.slow.cache.setSubsystems(snapshot) |
| 384 | if err != nil { |
| 385 | return fmt.Errorf("subsystems: %w", err) |
| 386 | } |
| 387 | return nil |
| 388 | }) |
| 389 | |
| 390 | if c.CollectPlanCacheMetrics.IsEnabled() { |
| 391 | group.Go(func() error { |
| 392 | snapshot, err := c.fetchPlanCache(groupCtx, beat, c.slowExec, c.slowDoQuery) |
| 393 | c.slow.cache.setPlanCache(snapshot) |
| 394 | if err != nil { |
| 395 | return fmt.Errorf("plan cache: %w", err) |
| 396 | } |
| 397 | return nil |
| 398 | }) |
| 399 | } else { |
| 400 | c.slow.cache.setPlanCache(planCacheSnapshot{ |
| 401 | timestamp: beat, |
| 402 | err: nil, |
| 403 | values: make(map[string]planCacheInstanceMetrics), |
| 404 | meta: make(map[string]planCacheMetrics), |
| 405 | }) |
| 406 | } |
| 407 | |
| 408 | if err := group.Wait(); err != nil && !errors.Is(err, context.Canceled) { |
| 409 | c.logErrorOnce("slow_path_error", "slow path: %s", trimDriverMessage(err)) |
| 410 | } else if err == nil { |
| 411 | c.clearErrorOnce("slow_path_error") |
| 412 | } |
| 413 | } |
| 414 | |
| 415 | type queryFunc func(ctx context.Context, queryName, query string, assign func(column, value string, lineEnd bool)) error |
| 416 | type queryRowFunc func(ctx context.Context, queryName, query string, assign func(column, value string)) error |
| 417 | type execFunc func(ctx context.Context, query string) error |
| 418 | |
| 419 | func (c *Collector) fetchMessageQueues(ctx context.Context, beat time.Time, do queryFunc) (messageQueueSnapshot, error) { |
| 420 | snapshot := messageQueueSnapshot{ |
| 421 | metrics: make(map[string]messageQueueInstanceMetrics), |
| 422 | meta: make(map[string]messageQueueMetrics), |
| 423 | timestamp: beat, |
| 424 | } |
| 425 | |
| 426 | if len(c.messageQueueTargets) == 0 { |
| 427 | return snapshot, nil |
| 428 | } |
| 429 | |
| 430 | var firstErr error |
| 431 | |
| 432 | for _, target := range c.messageQueueTargets { |
| 433 | key := target.ID() |
| 434 | errorKey := "slow_message_queue_" + key |
| 435 | meta := messageQueueMetrics{ |
| 436 | library: target.Library, |
| 437 | name: target.Name, |
| 438 | } |
| 439 | metrics := messageQueueInstanceMetrics{} |
| 440 | |
| 441 | queryName := fmt.Sprintf("message_queue_%s_%s", target.Library, target.Name) |
| 442 | query := buildMessageQueueQuery(target, c.supportsMessageQueueTableFunction()) |
| 443 | err := do(ctx, queryName, query, func(column, value string, lineEnd bool) { |
| 444 | switch column { |
| 445 | case "MESSAGE_COUNT": |
| 446 | metrics.Total = parseInt64OrZero(value) |
| 447 | case "INFORMATIONAL_MESSAGES": |
| 448 | metrics.Informational = parseInt64OrZero(value) |
| 449 | case "INQUIRY_MESSAGES": |
| 450 | metrics.Inquiry = parseInt64OrZero(value) |
| 451 | case "DIAGNOSTIC_MESSAGES": |
| 452 | metrics.Diagnostic = parseInt64OrZero(value) |
| 453 | case "ESCAPE_MESSAGES": |
| 454 | metrics.Escape = parseInt64OrZero(value) |
| 455 | case "NOTIFY_MESSAGES": |
| 456 | metrics.Notify = parseInt64OrZero(value) |
| 457 | case "SENDER_COPY_MESSAGES": |
| 458 | metrics.SenderCopy = parseInt64OrZero(value) |
| 459 | case "MAX_SEVERITY": |
| 460 | metrics.MaxSeverity = parseInt64OrZero(value) |
| 461 | } |
| 462 | }) |
| 463 | |
| 464 | if err != nil { |
| 465 | c.logQueryErrorOnce(errorKey, query, err) |
| 466 | if firstErr == nil { |
| 467 | firstErr = fmt.Errorf("message queue %s: %w", key, err) |
| 468 | } |
| 469 | continue |
| 470 | } |
| 471 | c.clearErrorOnce(errorKey) |
| 472 | |
| 473 | snapshot.metrics[key] = metrics |
| 474 | snapshot.meta[key] = meta |
| 475 | } |
| 476 | |
| 477 | snapshot.err = firstErr |
| 478 | return snapshot, firstErr |
| 479 | } |
| 480 | |
| 481 | func (c *Collector) fetchJobQueues(ctx context.Context, beat time.Time, do queryFunc) (jobQueueSnapshot, error) { |
| 482 | snapshot := jobQueueSnapshot{ |
| 483 | metrics: make(map[string]jobQueueInstanceMetrics), |
| 484 | meta: make(map[string]jobQueueMetrics), |
| 485 | timestamp: beat, |
| 486 | } |
| 487 | |
| 488 | if len(c.jobQueueTargets) == 0 { |
| 489 | return snapshot, nil |
| 490 | } |
| 491 | |
| 492 | var firstErr error |
| 493 | |
| 494 | for _, target := range c.jobQueueTargets { |
| 495 | key := target.ID() |
| 496 | errorKey := "slow_job_queue_" + key |
| 497 | meta := jobQueueMetrics{ |
| 498 | library: target.Library, |
| 499 | name: target.Name, |
| 500 | status: "UNKNOWN", |
| 501 | } |
| 502 | metrics := jobQueueInstanceMetrics{} |
| 503 | found := false |
| 504 | |
| 505 | queryName := fmt.Sprintf("job_queue_%s_%s", target.Library, target.Name) |
| 506 | query := buildJobQueueQuery(target) |
| 507 | err := do(ctx, queryName, query, func(column, value string, lineEnd bool) { |
| 508 | switch column { |
| 509 | case "JOB_QUEUE_STATUS": |
| 510 | meta.status = strings.TrimSpace(value) |
| 511 | case "NUMBER_OF_JOBS": |
| 512 | metrics.NumberOfJobs = parseInt64OrZero(value) |
| 513 | case "RELEASED_JOBS": |
| 514 | meta.jobsWaiting = parseInt64OrZero(value) |
| 515 | case "SCHEDULED_JOBS": |
| 516 | meta.jobsScheduled = parseInt64OrZero(value) |
| 517 | case "HELD_JOBS": |
| 518 | meta.jobsHeld = parseInt64OrZero(value) |
| 519 | case "MAXIMUM_ACTIVE_JOBS": |
| 520 | meta.maxJobs = parseInt64OrZero(value) |
| 521 | } |
| 522 | if lineEnd { |
| 523 | found = true |
| 524 | } |
| 525 | }) |
| 526 | |
| 527 | if err != nil { |
| 528 | c.logQueryErrorOnce(errorKey, query, err) |
| 529 | if firstErr == nil { |
| 530 | firstErr = fmt.Errorf("job queue %s: %w", key, err) |
| 531 | } |
| 532 | continue |
| 533 | } |
| 534 | c.clearErrorOnce(errorKey) |
| 535 | |
| 536 | if !found { |
| 537 | meta.status = "NOT_FOUND" |
| 538 | } |
| 539 | |
| 540 | snapshot.metrics[key] = metrics |
| 541 | snapshot.meta[key] = meta |
| 542 | } |
| 543 | |
| 544 | snapshot.err = firstErr |
| 545 | return snapshot, firstErr |
| 546 | } |
| 547 | |
| 548 | func (c *Collector) fetchOutputQueues(ctx context.Context, beat time.Time, do queryFunc) (outputQueueSnapshot, error) { |
| 549 | snapshot := outputQueueSnapshot{ |
| 550 | metrics: make(map[string]outputQueueInstanceMetrics), |
| 551 | meta: make(map[string]outputQueueMetrics), |
| 552 | timestamp: beat, |
| 553 | } |
| 554 | |
| 555 | if len(c.outputQueueTargets) == 0 { |
| 556 | return snapshot, nil |
| 557 | } |
| 558 | |
| 559 | var firstErr error |
| 560 | |
| 561 | for _, target := range c.outputQueueTargets { |
| 562 | key := target.ID() |
| 563 | errorEntriesKey := "slow_output_queue_entries_" + key |
| 564 | errorInfoKey := "slow_output_queue_info_" + key |
| 565 | meta := outputQueueMetrics{ |
| 566 | library: target.Library, |
| 567 | name: target.Name, |
| 568 | status: "UNKNOWN", |
| 569 | } |
| 570 | |
| 571 | metrics := outputQueueInstanceMetrics{} |
| 572 | entriesCount := int64(0) |
| 573 | entriesUsed := false |
| 574 | |
| 575 | queryName := fmt.Sprintf("output_queue_%s_%s", target.Library, target.Name) |
| 576 | entriesQuery := buildOutputQueueEntriesQuery(target) |
| 577 | err := do(ctx, queryName, entriesQuery, func(column, value string, lineEnd bool) { |
| 578 | if lineEnd { |
| 579 | entriesCount++ |
| 580 | } |
| 581 | }) |
| 582 | if err != nil { |
| 583 | c.logQueryErrorOnce(errorEntriesKey, entriesQuery, err) |
| 584 | if firstErr == nil { |
| 585 | firstErr = fmt.Errorf("output queue %s (entries): %w", key, err) |
| 586 | } |
| 587 | } else { |
| 588 | c.clearErrorOnce(errorEntriesKey) |
| 589 | entriesUsed = true |
| 590 | metrics.Files = entriesCount |
| 591 | } |
| 592 | |
| 593 | infoQuery := buildOutputQueueInfoQuery(target) |
| 594 | viewErr := do(ctx, queryName+"_view", infoQuery, func(column, value string, lineEnd bool) { |
| 595 | switch column { |
| 596 | case "OUTPUT_QUEUE_STATUS": |
| 597 | meta.status = strings.TrimSpace(value) |
| 598 | case "NUMBER_OF_WRITERS": |
| 599 | metrics.Writers = parseInt64OrZero(value) |
| 600 | case "NUMBER_OF_FILES": |
| 601 | if !entriesUsed { |
| 602 | metrics.Files = parseInt64OrZero(value) |
| 603 | } |
| 604 | } |
| 605 | }) |
| 606 | if viewErr != nil { |
| 607 | c.logQueryErrorOnce(errorInfoKey, infoQuery, viewErr) |
| 608 | if firstErr == nil { |
| 609 | firstErr = fmt.Errorf("output queue %s (info): %w", key, viewErr) |
| 610 | } |
| 611 | continue |
| 612 | } |
| 613 | c.clearErrorOnce(errorInfoKey) |
| 614 | |
| 615 | metrics.Released = boolToInt(strings.EqualFold(meta.status, "RELEASED")) |
| 616 | snapshot.metrics[key] = metrics |
| 617 | snapshot.meta[key] = meta |
| 618 | } |
| 619 | |
| 620 | snapshot.err = firstErr |
| 621 | return snapshot, firstErr |
| 622 | } |
| 623 | |
| 624 | func (c *Collector) countSubsystemsWith(doRow queryRowFunc, ctx context.Context) (int, error) { |
| 625 | var count int64 |
| 626 | err := doRow(ctx, "count_subsystems", queryCountSubsystems, func(column, value string) { |
| 627 | if column == "COUNT" { |
| 628 | if v, ok := c.parseInt64Value(value, 1); ok { |
| 629 | count = v |
| 630 | } |
| 631 | } |
| 632 | }) |
| 633 | return int(count), err |
| 634 | } |
| 635 | |
| 636 | func (c *Collector) fetchSubsystems(ctx context.Context, beat time.Time, do queryFunc, doRow queryRowFunc) (subsystemSnapshot, error) { |
| 637 | snapshot := subsystemSnapshot{ |
| 638 | metrics: make(map[string]subsystemInstanceMetrics), |
| 639 | meta: make(map[string]subsystemMetrics), |
| 640 | timestamp: beat, |
| 641 | } |
| 642 | |
| 643 | query := querySubsystems |
| 644 | if c.MaxSubsystems > 0 { |
| 645 | if total, err := c.countSubsystemsWith(doRow, ctx); err != nil { |
| 646 | c.logOnce("subsystem_count_failed", "failed to count subsystems before applying limit: %v", err) |
| 647 | } else if total > c.MaxSubsystems { |
| 648 | c.logOnce("subsystem_limit", "subsystem count (%d) exceeds limit (%d); truncating results", total, c.MaxSubsystems) |
| 649 | } |
| 650 | query = withFetchLimit(query, c.MaxSubsystems) |
| 651 | } |
| 652 | |
| 653 | currentSubsystem := "" |
| 654 | err := do(ctx, "subsystems", query, func(column, value string, lineEnd bool) { |
| 655 | switch column { |
| 656 | case "SUBSYSTEM_NAME": |
| 657 | name := strings.TrimSpace(value) |
| 658 | if name == "" { |
| 659 | currentSubsystem = "" |
| 660 | return |
| 661 | } |
| 662 | if c.subsystemSelector != nil && !c.subsystemSelector.MatchString(name) { |
| 663 | currentSubsystem = "" |
| 664 | return |
| 665 | } |
| 666 | currentSubsystem = name |
| 667 | subsystem := subsystemMetrics{name: name, status: "ACTIVE"} |
| 668 | parts := strings.SplitN(name, "/", 2) |
| 669 | if len(parts) == 2 { |
| 670 | subsystem.library = parts[0] |
| 671 | subsystem.name = parts[1] |
| 672 | } |
| 673 | snapshot.meta[currentSubsystem] = subsystem |
| 674 | case "CURRENT_ACTIVE_JOBS": |
| 675 | if currentSubsystem != "" { |
| 676 | if v, ok := c.parseInt64Value(value, 1); ok { |
| 677 | if metrics, exists := snapshot.metrics[currentSubsystem]; exists { |
| 678 | metrics.CurrentActiveJobs = v |
| 679 | snapshot.metrics[currentSubsystem] = metrics |
| 680 | } else { |
| 681 | snapshot.metrics[currentSubsystem] = subsystemInstanceMetrics{CurrentActiveJobs: v} |
| 682 | } |
| 683 | } |
| 684 | } |
| 685 | case "MAXIMUM_ACTIVE_JOBS": |
| 686 | if currentSubsystem != "" { |
| 687 | if v, ok := c.parseInt64Value(value, 1); ok { |
| 688 | if metrics, exists := snapshot.metrics[currentSubsystem]; exists { |
| 689 | metrics.MaximumActiveJobs = v |
| 690 | snapshot.metrics[currentSubsystem] = metrics |
| 691 | } else { |
| 692 | snapshot.metrics[currentSubsystem] = subsystemInstanceMetrics{MaximumActiveJobs: v} |
| 693 | } |
| 694 | } |
| 695 | } |
| 696 | } |
| 697 | |
| 698 | if lineEnd { |
| 699 | currentSubsystem = "" |
| 700 | } |
| 701 | }) |
| 702 | |
| 703 | if err != nil { |
| 704 | c.logQueryErrorOnce("slow_subsystems", query, err) |
| 705 | snapshot.err = err |
| 706 | return snapshot, err |
| 707 | } |
| 708 | c.clearErrorOnce("slow_subsystems") |
| 709 | snapshot.err = nil |
| 710 | return snapshot, nil |
| 711 | } |
| 712 | |
| 713 | func (c *Collector) fetchPlanCache(ctx context.Context, beat time.Time, exec execFunc, do queryFunc) (planCacheSnapshot, error) { |
| 714 | snapshot := planCacheSnapshot{ |
| 715 | values: make(map[string]planCacheInstanceMetrics), |
| 716 | meta: make(map[string]planCacheMetrics), |
| 717 | timestamp: beat, |
| 718 | } |
| 719 | |
| 720 | if err := exec(ctx, callAnalyzePlanCache); err != nil { |
| 721 | c.logQueryErrorOnce("slow_plan_cache_analyze", callAnalyzePlanCache, err) |
| 722 | snapshot.err = fmt.Errorf("analyze plan cache: %w", err) |
| 723 | return snapshot, snapshot.err |
| 724 | } |
| 725 | |
| 726 | var currentHeading string |
| 727 | err := do(ctx, "plan_cache_summary", queryPlanCacheSummary, func(column, value string, lineEnd bool) { |
| 728 | switch column { |
| 729 | case "HEADING": |
| 730 | currentHeading = strings.TrimSpace(value) |
| 731 | case "VALUE": |
| 732 | if currentHeading == "" { |
| 733 | return |
| 734 | } |
| 735 | key := planCacheMetricKey(currentHeading) |
| 736 | if key == "" { |
| 737 | return |
| 738 | } |
| 739 | if parsed, ok := c.parseInt64Value(value, precision); ok { |
| 740 | snapshot.values[key] = planCacheInstanceMetrics{Value: parsed} |
| 741 | snapshot.meta[key] = planCacheMetrics{heading: currentHeading} |
| 742 | } |
| 743 | } |
| 744 | if lineEnd { |
| 745 | currentHeading = "" |
| 746 | } |
| 747 | }) |
| 748 | |
| 749 | if err != nil { |
| 750 | c.logQueryErrorOnce("slow_plan_cache_summary", queryPlanCacheSummary, err) |
| 751 | snapshot.err = fmt.Errorf("plan cache summary: %w", err) |
| 752 | return snapshot, snapshot.err |
| 753 | } |
| 754 | c.clearErrorOnce("slow_plan_cache_summary") |
| 755 | |
| 756 | return snapshot, nil |
| 757 | } |
| 758 | |
| 759 | func (c *Collector) slowDoQuery(ctx context.Context, queryName, query string, assign func(column, value string, lineEnd bool)) error { |
| 760 | if c.slow.client == nil { |
| 761 | return errors.New("slow path client not initialised") |
| 762 | } |
| 763 | |
| 764 | start := time.Now() |
| 765 | err := c.queryWithClient(ctx, c.slow.client, queryName, query, assign) |
| 766 | elapsed := time.Since(start) |
| 767 | c.slow.cache.addLatency(queryName, elapsed.Microseconds()) |
| 768 | return err |
| 769 | } |
| 770 | |
| 771 | func (c *Collector) slowDoQueryRow(ctx context.Context, queryName, query string, assign func(column, value string)) error { |
| 772 | if c.slow.client == nil { |
| 773 | return errors.New("slow path client not initialised") |
| 774 | } |
| 775 | |
| 776 | start := time.Now() |
| 777 | err := c.queryRowWithClient(ctx, c.slow.client, queryName, query, assign) |
| 778 | elapsed := time.Since(start) |
| 779 | c.slow.cache.addLatency(queryName, elapsed.Microseconds()) |
| 780 | return err |
| 781 | } |
| 782 | |
| 783 | func (c *Collector) slowExec(ctx context.Context, query string) error { |
| 784 | if c.slow.client == nil { |
| 785 | return errors.New("slow path client not initialised") |
| 786 | } |
| 787 | start := time.Now() |
| 788 | err := c.execWithClient(ctx, c.slow.client, query) |
| 789 | elapsed := time.Since(start) |
| 790 | c.slow.cache.addLatency("analyze_plan_cache", elapsed.Microseconds()) |
| 791 | return err |
| 792 | } |
| 793 | |
| 794 | func (c *Collector) queryWithClient(ctx context.Context, client *as400proto.Client, queryName, query string, assign func(column, value string, lineEnd bool)) error { |
| 795 | return client.Query(ctx, query, func(columns []string, values []string) error { |
| 796 | for idx, col := range columns { |
| 797 | assign(col, values[idx], idx == len(columns)-1) |
| 798 | } |
| 799 | return nil |
| 800 | }) |
| 801 | } |
| 802 | |
| 803 | func (c *Collector) queryRowWithClient(ctx context.Context, client *as400proto.Client, queryName, query string, assign func(column, value string)) error { |
| 804 | return client.QueryWithLimit(ctx, query, 1, func(columns []string, values []string) error { |
| 805 | for idx, col := range columns { |
| 806 | assign(col, values[idx]) |
| 807 | } |
| 808 | return nil |
| 809 | }) |
| 810 | } |
| 811 | |
| 812 | func (c *Collector) execWithClient(ctx context.Context, client *as400proto.Client, query string) error { |
| 813 | if err := client.Connect(ctx); err != nil { |
| 814 | return err |
| 815 | } |
| 816 | return client.Exec(ctx, query) |
| 817 | } |