1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
|
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>Adding Observability to X-RAG</title>
<link rel="shortcut icon" type="image/gif" href="/favicon.ico" />
<link rel="stylesheet" href="../style.css" />
<link rel="stylesheet" href="style-override.css" />
</head>
<body>
<p class="header">
<a href="https://foo.zone">Home</a> | <a href="https://codeberg.org/snonux/foo.zone/src/branch/content-md/gemfeed/DRAFT-x-rag-observability-hackathon.md">Markdown</a> | <a href="gemini://foo.zone/gemfeed/DRAFT-x-rag-observability-hackathon.gmi">Gemini</a>
</p>
<h1 style='display: inline' id='adding-observability-to-x-rag'>Adding Observability to X-RAG</h1><br />
<br />
<span>This blog post describes my hackathon efforts adding observability to X-RAG, a distributed Retrieval-Augmented Generation (RAG) platform built by my brother Florian. I especially made time available over the weekend to join his 3-day hackathon (attending 2 days) with the goal of instrumenting his existing distributed system with observability. What started as "let's add some metrics" turned into a comprehensive implementation of the three pillars of observability: tracing, metrics, and logs.</span><br />
<br />
<a class='textlink' href='https://github.com/florianbuetow/x-rag'>X-RAG source code on GitHub</a><br />
<br />
<h2 style='display: inline' id='table-of-contents'>Table of Contents</h2><br />
<br />
<ul>
<li><a href='#adding-observability-to-x-rag'>Adding Observability to X-RAG</a></li>
<li>⇢ <a href='#what-is-x-rag'>What is X-RAG?</a></li>
<li>⇢ <a href='#running-kubernetes-locally-with-kind'>Running Kubernetes locally with Kind</a></li>
<li>⇢ <a href='#motivation'>Motivation</a></li>
<li>⇢ <a href='#the-observability-stack'>The observability stack</a></li>
<li>⇢ <a href='#grafana-alloy-the-unified-collector'>Grafana Alloy: the unified collector</a></li>
<li>⇢ <a href='#centralised-logging-with-loki'>Centralised logging with Loki</a></li>
<li>⇢ ⇢ <a href='#alloy-configuration-for-logs'>Alloy configuration for logs</a></li>
<li>⇢ ⇢ <a href='#querying-logs-with-logql'>Querying logs with LogQL</a></li>
<li>⇢ <a href='#metrics-with-prometheus'>Metrics with Prometheus</a></li>
<li>⇢ ⇢ <a href='#alloy-configuration-for-application-metrics'>Alloy configuration for application metrics</a></li>
<li>⇢ ⇢ <a href='#kubernetes-metrics-kubelet-cadvisor-and-kube-state-metrics'>Kubernetes metrics: kubelet, cAdvisor, and kube-state-metrics</a></li>
<li>⇢ ⇢ <a href='#infrastructure-metrics-kafka-redis-minio'>Infrastructure metrics: Kafka, Redis, MinIO</a></li>
<li>⇢ <a href='#distributed-tracing-with-tempo'>Distributed tracing with Tempo</a></li>
<li>⇢ ⇢ <a href='#understanding-traces-spans-and-the-trace-tree'>Understanding traces, spans, and the trace tree</a></li>
<li>⇢ ⇢ <a href='#how-trace-context-propagates'>How trace context propagates</a></li>
<li>⇢ ⇢ <a href='#implementation'>Implementation</a></li>
<li>⇢ ⇢ <a href='#alloy-configuration-for-traces'>Alloy configuration for traces</a></li>
<li>⇢ <a href='#async-ingestion-trace-walkthrough'>Async ingestion trace walkthrough</a></li>
<li>⇢ ⇢ <a href='#step-1-ingest-a-document'>Step 1: Ingest a document</a></li>
<li>⇢ ⇢ <a href='#step-2-find-the-ingestion-trace'>Step 2: Find the ingestion trace</a></li>
<li>⇢ ⇢ <a href='#step-3-fetch-the-complete-trace'>Step 3: Fetch the complete trace</a></li>
<li>⇢ ⇢ <a href='#step-4-analyse-the-async-trace'>Step 4: Analyse the async trace</a></li>
<li>⇢ ⇢ <a href='#viewing-traces-in-grafana'>Viewing traces in Grafana</a></li>
<li>⇢ <a href='#end-to-end-search-trace-walkthrough'>End-to-end search trace walkthrough</a></li>
<li>⇢ ⇢ <a href='#step-1-make-a-search-request'>Step 1: Make a search request</a></li>
<li>⇢ ⇢ <a href='#step-2-query-tempo-for-the-trace'>Step 2: Query Tempo for the trace</a></li>
<li>⇢ ⇢ <a href='#step-3-analyse-the-trace'>Step 3: Analyse the trace</a></li>
<li>⇢ ⇢ <a href='#step-4-search-traces-with-traceql'>Step 4: Search traces with TraceQL</a></li>
<li>⇢ ⇢ <a href='#viewing-the-search-trace-in-grafana'>Viewing the search trace in Grafana</a></li>
<li>⇢ <a href='#correlating-the-three-signals'>Correlating the three signals</a></li>
<li>⇢ <a href='#grafana-dashboards'>Grafana dashboards</a></li>
<li>⇢ <a href='#results-two-days-well-spent'>Results: two days well spent</a></li>
<li>⇢ <a href='#slis-slos-and-slas'>SLIs, SLOs and SLAs</a></li>
<li>⇢ <a href='#using-amp-for-ai-assisted-development'>Using Amp for AI-assisted development</a></li>
<li>⇢ <a href='#other-changes-along-the-way'>Other changes along the way</a></li>
<li>⇢ <a href='#lessons-learned'>Lessons learned</a></li>
</ul><br />
<h2 style='display: inline' id='what-is-x-rag'>What is X-RAG?</h2><br />
<br />
<span>X-RAG is a distributed RAG (Retrieval-Augmented Generation) platform running on Kubernetes. The idea behind RAG is simple: instead of asking an LLM to answer questions from its training data alone, you first retrieve relevant documents from your own knowledge base, then feed those documents to the LLM as context. The LLM synthesises an answer grounded in your actual content—reducing hallucinations and enabling answers about private or recent information the model was never trained on.</span><br />
<br />
<span>X-RAG handles the full pipeline: ingest documents, chunk them into searchable pieces, generate vector embeddings, store them in a vector database, and at query time, retrieve relevant chunks and pass them to an LLM for answer generation. The system supports both local LLMs (Florian runs his on a beefy desktop) and cloud APIs like OpenAI. I configured an OpenAI API key since my laptop's CPU and GPU aren't fast enough for decent local inference.</span><br />
<br />
<span>All services are implemented in Python. I'm more used to Ruby, Go, and Bash these days, but for this project it didn't matter—Python's OpenTelemetry integration is straightforward, I wasn't planning to write or rewrite tons of application code, and with GenAI assistance the language barrier was a non-issue. The OpenTelemetry concepts and patterns should translate to other languages too—the SDK APIs are intentionally similar across Python, Go, Java, and others.</span><br />
<br />
<span>X-RAG consists of several independently scalable microservices:</span><br />
<br />
<ul>
<li>Search UI: FastAPI web interface for queries</li>
<li>Ingestion API: Document upload endpoint</li>
<li>Embedding Service: gRPC service for vector embeddings</li>
<li>Indexer: Kafka consumer that processes documents</li>
<li>Search Service: gRPC service orchestrating the RAG pipeline</li>
</ul><br />
<span>The Embedding Service deserves extra explanation because in the beginning I didn't really knew what it was. Text isn't directly searchable in a vector database—you need to convert it to numerical vectors (embeddings) that capture semantic meaning. The Embedding Service takes text chunks and calls an embedding model (OpenAI's <span class='inlinecode'>text-embedding-3-small</span> in my case, or a local model on Florian's setup) to produce these vectors. For the LLM search completion answer, I used <span class='inlinecode'>gpt-4o-mini</span>.</span><br />
<br />
<span>Similar concepts end up with similar vectors, so "What is machine learning?" and "Explain ML" produce vectors close together in the embedding space. At query time, your question gets embedded too, and the vector database finds chunks with nearby vectors—that's semantic search.</span><br />
<br />
<span>The data layer includes Weaviate (vector database with hybrid search), Kafka (message queue), MinIO (object storage), and Redis (cache). All of this runs in a Kind Kubernetes cluster for local development, with the same manifests deployable to production.</span><br />
<br />
<pre>
┌─────────────────────────────────────────────────────────────────────────┐
│ X-RAG Kubernetes Cluster │
├─────────────────────────────────────────────────────────────────────────┤
│ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │
│ │ Search UI │ │Search Svc │ │Embed Service│ │ Indexer │ │
│ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ │
│ │ │ │ │ │
│ └────────────────┴────────────────┴────────────────┘ │
│ │ │
│ ▼ │
│ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │
│ │ Weaviate │ │ Kafka │ │ MinIO │ │
│ └─────────────┘ └─────────────┘ └─────────────┘ │
└─────────────────────────────────────────────────────────────────────────┘
</pre>
<br />
<h2 style='display: inline' id='running-kubernetes-locally-with-kind'>Running Kubernetes locally with Kind</h2><br />
<br />
<span>X-RAG runs on Kubernetes, but you don't need a cloud account to develop it. The project uses Kind (Kubernetes in Docker)—a tool originally created by the Kubernetes SIG for testing Kubernetes itself.</span><br />
<br />
<a class='textlink' href='https://kind.sigs.k8s.io/'>Kind - Kubernetes in Docker</a><br />
<br />
<span>Kind spins up a full Kubernetes cluster using Docker containers as nodes. The control plane (API server, etcd, scheduler, controller-manager) runs in one container, and worker nodes run in separate containers. Inside these "node containers," pods run just like they would on real servers—using containerd as the container runtime. It's containers all the way down.</span><br />
<br />
<span>Technically, each Kind node is a Docker container running a minimal Linux image with kubelet and containerd installed. When you deploy a pod, kubelet inside the node container instructs containerd to pull and run the container image. So you have Docker running node containers, and inside those, containerd running application containers. Network-wise, Kind sets up a Docker bridge network and uses CNI plugins (kindnet by default) for pod networking within the cluster.</span><br />
<br />
<pre>
$ docker ps --format "table {{.Names}}\t{{.Image}}"
NAMES IMAGE
xrag-k8-control-plane kindest/node:v1.32.0
xrag-k8-worker kindest/node:v1.32.0
xrag-k8-worker2 kindest/node:v1.32.0
</pre>
<br />
<span>The <span class='inlinecode'>kindest/node</span> image contains everything needed: kubelet, containerd, CNI plugins, and pre-pulled pause containers. Port mappings in the Kind config expose services to the host—that's how http://localhost:8080 reaches the search-ui running inside a pod, inside a worker container, inside Docker.</span><br />
<br />
<pre>
┌─────────────────────────────────────────────────────────────────────────┐
│ Docker Host │
├─────────────────────────────────────────────────────────────────────────┤
│ ┌───────────────────┐ ┌───────────────────┐ ┌───────────────────┐ │
│ │ xrag-k8-control │ │ xrag-k8-worker │ │ xrag-k8-worker2 │ │
│ │ -plane (container)│ │ (container) │ │ (container) │ │
│ │ │ │ │ │ │ │
│ │ K8s API server │ │ Pods: │ │ Pods: │ │
│ │ etcd, scheduler │ │ • search-ui │ │ • weaviate │ │
│ │ │ │ • search-service │ │ • kafka │ │
│ │ │ │ • embedding-svc │ │ • prometheus │ │
│ │ │ │ • indexer │ │ • grafana │ │
│ └───────────────────┘ └───────────────────┘ └───────────────────┘ │
└─────────────────────────────────────────────────────────────────────────┘
</pre>
<br />
<span>Why Kind? It gives you a real Kubernetes environment—the same manifests deploy to production clouds unchanged. No minikube quirks, no Docker Compose translation layer. Just Kubernetes. I already have a k3s cluster running at home, but Kind made collaboration easier—everyone working on X-RAG gets the exact same setup by cloning the repo and running <span class='inlinecode'>make cluster-start</span>.</span><br />
<br />
<span>Florian developed X-RAG on macOS, but it worked seamlessly on my Linux laptop. The only difference was Docker's resource allocation: on macOS you configure limits in Docker Desktop, on Linux it uses host resources directly. That's because under macOS the Linux Docker containers run on an emulation layer as macOS is not Linux.</span><br />
<br />
<span>My hardware: a ThinkPad X1 Carbon Gen 9 with an 11th Gen Intel Core i7-1185G7 (4 cores, 8 threads at 3.00GHz) and 32GB RAM (running Fedora Linux). During the hackathon, memory usage peaked around 15GB—comfortable headroom. CPU was the bottleneck; with ~38 pods running across all namespaces (rag-system, monitoring, kube-system, etc.), plus Discord for the remote video call and Tidal streaming hi-res music, things got tight. When rebuilding Docker images or restarting the cluster, Discord video and audio would stutter—my fellow hackers probably wondered why I kept freezing mid-sentence. A beefier CPU would have meant less waiting and smoother calls, but it was manageable.</span><br />
<br />
<h2 style='display: inline' id='motivation'>Motivation</h2><br />
<br />
<span>When I joined the hackathon, Florian's X-RAG was functional but opaque. With five services communicating via gRPC, Kafka, and HTTP, debugging was cumbersome. When a search request take 5 seconds, there was no visibility into where the time was being spent. Was it the embedding generation? The vector search? The LLM synthesis? Nobody would be able to figure it out quickly.</span><br />
<br />
<span>Distributed systems are inherently opaque. Each service logs its own view of the world, but correlating events across service boundaries is archaeology. Grepping through logs on many pods, trying to mentally reconstruct what happened—not fun. This was the perfect hackathon project: Explore this Observability Stack in greater depth.</span><br />
<br />
<h2 style='display: inline' id='the-observability-stack'>The observability stack</h2><br />
<br />
<span>Before diving into implementation, here's what I deployed. The complete stack runs in the monitoring namespace:</span><br />
<br />
<pre>
$ kubectl get pods -n monitoring
NAME READY STATUS
alloy-84ddf4cd8c-7phjp 1/1 Running
grafana-6fcc89b4d6-pnh8l 1/1 Running
kube-state-metrics-5d954c569f-2r45n 1/1 Running
loki-8c9bbf744-sc2p5 1/1 Running
node-exporter-kb8zz 1/1 Running
node-exporter-zcrdz 1/1 Running
node-exporter-zmskc 1/1 Running
prometheus-7f755f675-dqcht 1/1 Running
tempo-55df7dbcdd-t8fg9 1/1 Running
</pre>
<br />
<span>Each component has a specific role:</span><br />
<br />
<ul>
<li><span class='inlinecode'>Grafana Alloy</span>: The unified collector. Receives OTLP from applications, scrapes Prometheus endpoints, tails log files. Think of it as the central nervous system.</li>
<li><span class='inlinecode'>Prometheus</span>: Time-series database for metrics. Stores counters, gauges, and histograms with 15-day retention.</li>
<li><span class='inlinecode'>Tempo</span>: Trace storage. Receives spans via OTLP, correlates them by trace ID, enables TraceQL queries.</li>
<li><span class='inlinecode'>Loki</span>: Log aggregation. Indexes labels (namespace, pod, container), stores log chunks, enables LogQL queries.</li>
<li><span class='inlinecode'>Grafana</span>: The unified UI. Queries all three backends, correlates signals, displays dashboards.</li>
<li><span class='inlinecode'>kube-state-metrics</span>: Exposes Kubernetes object metrics (pod status, deployments, resource requests).</li>
<li><span class='inlinecode'>node-exporter</span>: Exposes host-level metrics (CPU, memory, disk, network) from each Kubernetes node.</li>
</ul><br />
<span>Everything is accessible via port-forwards:</span><br />
<br />
<ul>
<li>Grafana: http://localhost:3000 (unified UI for all three signals)</li>
<li>Prometheus: http://localhost:9090 (metrics queries)</li>
<li>Tempo: http://localhost:3200 (trace queries)</li>
<li>Loki: http://localhost:3100 (log queries)</li>
</ul><br />
<h2 style='display: inline' id='grafana-alloy-the-unified-collector'>Grafana Alloy: the unified collector</h2><br />
<br />
<span>Before diving into the individual signals, I want to highlight Grafana Alloy—the component that ties everything together. Alloy is Grafana's vendor-neutral OpenTelemetry Collector distribution, and it became the backbone of the observability stack.</span><br />
<br />
<a class='textlink' href='https://grafana.com/docs/alloy/latest/'>Grafana Alloy documentation</a><br />
<br />
<span>Why use a centralised collector instead of having each service push directly to backends?</span><br />
<br />
<ul>
<li><span class='inlinecode'>Decoupling</span>: Applications don't need to know about Prometheus, Tempo, or Loki. They speak OTLP, and Alloy handles the translation.</li>
<li><span class='inlinecode'>Unified timestamps</span>: All telemetry flows through one system, making correlation in Grafana more reliable.</li>
<li><span class='inlinecode'>Processing pipeline</span>: Batch data before sending, filter noisy metrics, enrich with labels—all in one place.</li>
<li><span class='inlinecode'>Backend flexibility</span>: Switch from Tempo to Jaeger without changing application code.</li>
</ul><br />
<span>Alloy uses a configuration language called River, which feels similar to Terraform's HCL—declarative blocks with attributes. If you've written Terraform, River will look familiar. The full Alloy configuration runs to over 1400 lines with comments explaining each section. It handles OTLP receiving, batch processing, Prometheus export, Tempo export, Kubernetes metrics scraping, infrastructure metrics, and pod log collection. All three signals—metrics, traces, logs—flow through this single component, making Alloy the central nervous system of the observability stack.</span><br />
<br />
<span>In the following sections, I'll cover each observability pillar and show the relevant Alloy configuration for each.</span><br />
<br />
<h2 style='display: inline' id='centralised-logging-with-loki'>Centralised logging with Loki</h2><br />
<br />
<span>Getting all logs in one place was the foundation. I deployed Grafana Loki in the monitoring namespace, with Grafana Alloy running as a DaemonSet on each node to collect logs.</span><br />
<br />
<pre>
┌──────────────────────────────────────────────────────────────────────┐
│ LOGS PIPELINE │
├──────────────────────────────────────────────────────────────────────┤
│ Applications write to stdout → containerd stores in /var/log/pods │
│ │ │
│ File tail │
│ ▼ │
│ Grafana Alloy (DaemonSet) │
│ Discovers pods, extracts metadata │
│ │ │
│ HTTP POST /loki/api/v1/push │
│ ▼ │
│ Grafana Loki │
│ Indexes labels, stores chunks │
└──────────────────────────────────────────────────────────────────────┘
</pre>
<br />
<h3 style='display: inline' id='alloy-configuration-for-logs'>Alloy configuration for logs</h3><br />
<br />
<span>Alloy discovers pods via the Kubernetes API, tails their log files from /var/log/pods/, and ships to Loki. Importantly, Alloy runs as a DaemonSet on each worker node—it doesn't run inside the application pods. Since containerd writes all container stdout/stderr to /var/log/pods/ on the node's filesystem, Alloy can tail logs for every pod on that node from a single location without any sidecar injection:</span><br />
<br />
<pre>
loki.source.kubernetes "pod_logs" {
targets = discovery.relabel.pod_logs.output
forward_to = [loki.process.pod_logs.receiver]
}
loki.write "default" {
endpoint {
url = "http://loki.monitoring.svc.cluster.local:3100/loki/api/v1/push"
}
}
</pre>
<br />
<h3 style='display: inline' id='querying-logs-with-logql'>Querying logs with LogQL</h3><br />
<br />
<span>Now I could query logs in Loki (e.g. via Grafana UI) with LogQL:</span><br />
<br />
<pre>
{namespace="rag-system", container="search-ui"} |= "ERROR"
</pre>
<br />
<h2 style='display: inline' id='metrics-with-prometheus'>Metrics with Prometheus</h2><br />
<br />
<span>I added Prometheus metrics to every service. Following the Four Golden Signals (latency, traffic, errors, saturation), I instrumented the codebase with histograms, counters, and gauges:</span><br />
<br />
<!-- Generator: GNU source-highlight 3.1.9
by Lorenzo Bettini
http://www.lorenzobettini.it
http://www.gnu.org/software/src-highlite -->
<pre><b><u><font color="#000000">from</font></u></b> prometheus_client <b><u><font color="#000000">import</font></u></b> Histogram, Counter, Gauge
search_duration = Histogram(
<font color="#808080">"search_service_request_duration_seconds"</font>,
<font color="#808080">"Total duration of Search Service requests"</font>,
[<font color="#808080">"method"</font>],
buckets=[<font color="#000000">0.1</font>, <font color="#000000">0.25</font>, <font color="#000000">0.5</font>, <font color="#000000">1.0</font>, <font color="#000000">2.5</font>, <font color="#000000">5.0</font>, <font color="#000000">10.0</font>, <font color="#000000">20.0</font>, <font color="#000000">30.0</font>, <font color="#000000">60.0</font>],
)
errors_total = Counter(
<font color="#808080">"search_service_errors_total"</font>,
<font color="#808080">"Error count by type"</font>,
[<font color="#808080">"method"</font>, <font color="#808080">"error_type"</font>],
)
</pre>
<br />
<span>Initially, I used Prometheus scraping—each service exposed a /metrics endpoint, and Prometheus pulled metrics every 15 seconds. This worked, but I wanted a unified pipeline.</span><br />
<br />
<h3 style='display: inline' id='alloy-configuration-for-application-metrics'>Alloy configuration for application metrics</h3><br />
<br />
<span>The breakthrough came with Grafana Alloy as an OpenTelemetry collector. Services now push metrics via OTLP (OpenTelemetry Protocol), and Alloy converts them to Prometheus format:</span><br />
<br />
<pre>
┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
│ search-ui │ │search-svc │ │embed-svc │ │ indexer │
│ OTel Meter │ │ OTel Meter │ │ OTel Meter │ │ OTel Meter │
│ │ │ │ │ │ │ │ │ │ │ │
│ OTLPExporter│ │ OTLPExporter│ │ OTLPExporter│ │ OTLPExporter│
└──────┬──────┘ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘
│ │ │ │
└────────────────┴────────────────┴────────────────┘
│
▼ OTLP/gRPC (port 4317)
┌─────────────────────┐
│ Grafana Alloy │
└──────────┬──────────┘
│ prometheus.remote_write
▼
┌─────────────────────┐
│ Prometheus │
└─────────────────────┘
</pre>
<br />
<span>Alloy receives OTLP on ports 4317 (gRPC) or 4318 (HTTP), batches the data for efficiency, and exports to Prometheus:</span><br />
<br />
<pre>
otelcol.receiver.otlp "default" {
|
