Conversation
Evolve the KloudMate Agent section for agent v2: - Rewrite index.mdx around the "install once, instrument everything" model, the L0–L5 layered stack, and a platform capability matrix. - Add concepts/: architecture (control plane vs data plane), layered instrumentation, discovery, and the managed/manual configuration model. - Port the architecture, instrumentation-flow, and benefits SVG diagrams. - Scaffold the remaining v2 IA as draft stubs so cross-wave links resolve; set target sidebar order on ebpf-observability (5) and management (11). Sources: docs/agentv2 HLD 00/01/02, lld-discovery, lld-config-model, ADR-0003, and user-facing-feature-inventory. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Add baseline/host-metrics-and-logs (Layer 1) with per-mode coverage (hostmetrics/journald/filelog, Windows perf counters/Event Log, kubeletstats/k8s_cluster, docker stats) and a "Zero-config baseline" group. - Rewrite ebpf-observability for the v2 default-on model: kernel matrix (4.14/5.8+BTF/5.17), capability gating, graceful degradation, overhead budget, and tuning knobs; preserve the manual-mode YAML. - Update installation pages (linux, docker, kubernetes, windows, ASG, IaC) with managed-mode notes, a post-install checklist, and mode-specific privilege/coverage pointers; no install commands changed. Sources: docs/agentv2 HLD 03, lld-ebpf-enablement, ADR-0005, lld-infra-receivers, 01-current-state, components_windows. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add the "Application APM" section (L3-L5, opt-in per service): - overview: per-service Instrument flow, status chips, restart consent + health gate, automatic crash-loop rollback, clean-trace suppression, and the autonomous "instrument everything" advanced note. - linux-systemd: systemd drop-ins and per-language env (Java/Node/Python/.NET), bundled lockstep assets, restart orchestration, non-systemd caveat. - kubernetes: OpenTelemetry Operator annotation injection, parity with Linux. - windows-dotnet: CLR profiler via registry/applicationHost.config, IIS per-pool vs 8.5 host-wide, Windows services, Java/Node/Python registry env, and the ETW HTTP.sys baseline. - php: version-gated native (8.2+) vs dd-trace (7.0-8.1), overhead, trace parity, container path. - go-ebpf: eBPF-only coverage and cross-service parent-span correlation. - sampling: global default + per-service override, parent-based sampling. Sources: docs/agentv2 HLD 04/06/07/09, ADR-0001/0002/0004, lld-vm-instrumentation, lld-control-plane-ui, lld-php-graduation, lld-windows-etwreceiver, lld-ebpf-sdk-correlation. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add the "Database monitoring" section plus web-server monitoring: - overview: eBPF wire-side DAM vs server-side receivers, the profile/coverage concept, the discovery-led wizard, and DryRun-gated safety. - engines-and-coverages: the actual engines and coverages from internal/dbmonitor/profiles (Postgres/MySQL/SQL Server/Oracle/Mongo/Redis/ Elasticsearch/ClickHouse/SAP HANA/Snowflake), connection- vs file-based coverages, and the km.db.coverage log tag. - on-vm: endpoint canonicalization and file-based coverages. - on-kubernetes: singleton Deployment collector, Service-DNS discovery prefill, managed-mode requirement, the KM_SECRET_/dbMonitoring.secretName Secret path, and the Postgres worked example (live-validated §11). - credentials: env-ref (default), plaintext inline, operator-managed, rotation. - web-server-monitoring: nginx/apache status-module probing, read-only guarantee. Sources: docs/agentv2 lld-db-monitoring, lld-db-monitoring-k8s, lld-web-server-monitoring, and internal/dbmonitor/profiles. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- self-observability: Tier 1 out-of-band kmagent.* health metrics (catalog, collector.up reasons, dead-man's-switch), Tier 2 opt-in otelcol_* self- telemetry, and out-of-band ERROR self-logs. - platform-notes: Kubernetes (fleet manager + DaemonSet/Deployment, unified protocol, real per-node collector status, self-logs, ConfigMap ownership), Windows (IIS/SCM/WMI discovery, ETW baseline, injection, Local System), Docker (infra + eBPF today, in-place PHP, inject-spec + planned OCI-hook, deployment posture). - management: refresh for managed vs manual mode, config history/rollback, agent health/self-logs status, and agent updates. Sources: docs/agentv2 lld-agent-health, ADR-0006, hld/08-docker-on-vm, lld-container-instrumentation, hld/09, roadmap Phase 7. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- reference: metric catalogs (host, eBPF, per-engine DB, kmagent.*), telemetry identity (db.system.name values, server.address/port, service.name = engine:addr:port, dashboards keyed on server address), design decisions (ADR-0001..0006 in user terms), and a glossary. - advanced-configuration: manual YAML, custom receivers, sampling, eBPF overhead tuning, self-telemetry, autonomous mode, credential rotation. - troubleshooting: symptom-first, using health metrics and self-logs as the spine, across install, metrics, traces, databases, Windows, and Kubernetes. - cross-links: reciprocal pointers from apm-and-tracing, database-monitoring, infrastructure, and opentelemetry into the agent section. All 39 agent pages now publish; build and link check pass with no drafts. Sources: docs/agentv2 lld-db-monitoring (compile.go/identity.go), lld-agent-health, lld-config-model, decisions/ADR-0001..0006, internal/dbmonitor/metricmeta. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Editorial pass across the agent docs per review: - Remove all "agent v2" / "v2" wording; the docs describe "the agent". - Simplify language and reduce internal detail: drop deep mechanism (systemd drop-in paths, receiver/pipeline/compile internals, the in-process-collector explanation) in favour of what users see and do. - Cut the database connection-based vs file-based coverage framing; focus on what can be monitored, not how. - Remove AI-sounding phrasing (blast radius, escape hatch, wire-side, golden-signal, "compose rather than overlap", etc.) and any em dashes. - Drop the internal architecture diagram and the marketing benefits graphic; clean "v2"/em dashes out of the instrumentation-flow diagram. Build, lint, and link check all pass (258 pages, no broken links). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Review and content update Agent V2 pages
Replace the GCP egress IPs with the new AWS Elastic IPs (Asia/Europe/US) as the synthetic workers migrate from GCP to AWS. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Fills in the Profiling overview page and adds a page per view (All Services, Profile Types, Labels, Flame Graph, Diff), plus a Getting Started section covering Pyroscope SDK setup with Basic auth against the KloudMate ingest endpoint. Also fixes validate-frontmatter.mjs, which resolved its target directory via URL.pathname and broke on Windows (doubled drive letter) instead of using fileURLToPath. Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
…e graph pages Prose-only pages left the reader guessing what a card or the flame graph actually looks like. Adds a screenshot after each page's "reading a card"/"the flame graph" section.
"Diff" wasn't a declarative route name; renames diff.mdx to difference-view.mdx (title and route both) and embeds a screenshot of the merged/side-by-side toggle.
Go/Java/Python/Node.js/Ruby/.NET/Rust setup steps lived as inline tabs on the profiling overview page, growing that page past what an overview needs and giving each language no page of its own to deep-link into. Moves them to guides/profiling/ (one page per language, plus an eBPF whole-host profiling option for anything not on the list), matching how guides/opentelemetry/ is already organized. The overview page now links out to Guides instead of hosting the setup content directly, and its ingest-endpoint and intro copy is brought up to date.
…and fix outdated content
…and fix outdated content
Refresh and update screenshots in Alerts. Reports and API monitoring Docs.
Docs/profiling
Added Screenshots in agents v2 and added Network requirements
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.