diff options
| -rw-r--r-- | gemfeed/2025-12-07-f3s-kubernetes-with-freebsd-part-8.html | 1037 | ||||
| -rw-r--r-- | gemfeed/DRAFT-f3s-kubernetes-with-freebsd-part-8b.html | 1164 | ||||
| -rw-r--r-- | gemfeed/atom.xml | 1313 | ||||
| -rw-r--r-- | gemfeed/f3s-kubernetes-with-freebsd-part-8/grafana-etcd-dashboard.png | bin | 0 -> 201310 bytes | |||
| -rw-r--r-- | gemfeed/f3s-kubernetes-with-freebsd-part-8/grafana-zfs-arc-stats.png | bin | 0 -> 168537 bytes | |||
| -rw-r--r-- | gemfeed/f3s-kubernetes-with-freebsd-part-8/grafana-zfs-dashboard.png | bin | 0 -> 210342 bytes | |||
| -rw-r--r-- | gemfeed/f3s-kubernetes-with-freebsd-part-8/grafana-zfs-datasets.png | bin | 0 -> 149339 bytes |
7 files changed, 2192 insertions, 1322 deletions
diff --git a/gemfeed/2025-12-07-f3s-kubernetes-with-freebsd-part-8.html b/gemfeed/2025-12-07-f3s-kubernetes-with-freebsd-part-8.html index 53bc16d9..0cda1b53 100644 --- a/gemfeed/2025-12-07-f3s-kubernetes-with-freebsd-part-8.html +++ b/gemfeed/2025-12-07-f3s-kubernetes-with-freebsd-part-8.html @@ -18,7 +18,7 @@ </p> <h1 style='display: inline' id='f3s-kubernetes-with-freebsd---part-8-observability'>f3s: Kubernetes with FreeBSD - Part 8: Observability</h1><br /> <br /> -<span class='quote'>Published at 2025-12-06T23:58:24+02:00</span><br /> +<span class='quote'>Published at 2025-12-06T23:58:24+02:00, last updated Mon 09 Mar 09:33:08 EET 2026</span><br /> <br /> <span>This is the 8th blog post about the f3s series for my self-hosting demands in a home lab. f3s? The "f" stands for FreeBSD, and the "3s" stands for k3s, the Kubernetes distribution I use on FreeBSD-based physical machines.</span><br /> <br /> @@ -60,23 +60,56 @@ <li>⇢ ⇢ <a href='#adding-freebsd-hosts-to-prometheus'>Adding FreeBSD hosts to Prometheus</a></li> <li>⇢ ⇢ <a href='#freebsd-memory-metrics-compatibility'>FreeBSD memory metrics compatibility</a></li> <li>⇢ ⇢ <a href='#disk-io-metrics-limitation'>Disk I/O metrics limitation</a></li> +<li>⇢ <a href='#zfs-monitoring-for-freebsd-servers'>ZFS Monitoring for FreeBSD Servers</a></li> +<li>⇢ ⇢ <a href='#node-exporter-zfs-collector'>Node Exporter ZFS Collector</a></li> +<li>⇢ ⇢ <a href='#verifying-zfs-metrics'>Verifying ZFS Metrics</a></li> +<li>⇢ ⇢ <a href='#zfs-recording-rules'>ZFS Recording Rules</a></li> +<li>⇢ ⇢ <a href='#grafana-dashboards'>Grafana Dashboards</a></li> +<li>⇢ ⇢ <a href='#deployment'>Deployment</a></li> +<li>⇢ ⇢ <a href='#verifying-zfs-metrics-in-prometheus'>Verifying ZFS Metrics in Prometheus</a></li> +<li>⇢ ⇢ <a href='#key-metrics-to-monitor'>Key Metrics to Monitor</a></li> +<li>⇢ ⇢ <a href='#zfs-pool-and-dataset-metrics-via-textfile-collector'>ZFS Pool and Dataset Metrics via Textfile Collector</a></li> <li>⇢ <a href='#monitoring-external-openbsd-hosts'>Monitoring external OpenBSD hosts</a></li> <li>⇢ ⇢ <a href='#installing-node-exporter-on-openbsd'>Installing Node Exporter on OpenBSD</a></li> <li>⇢ ⇢ <a href='#adding-openbsd-hosts-to-prometheus'>Adding OpenBSD hosts to Prometheus</a></li> <li>⇢ ⇢ <a href='#openbsd-memory-metrics-compatibility'>OpenBSD memory metrics compatibility</a></li> +<li>⇢ <a href='#distributed-tracing-with-grafana-tempo'>Distributed Tracing with Grafana Tempo</a></li> +<li>⇢ ⇢ <a href='#why-distributed-tracing'>Why Distributed Tracing?</a></li> +<li>⇢ ⇢ <a href='#deploying-grafana-tempo'>Deploying Grafana Tempo</a></li> +<li>⇢ <a href='#-configuration-strategy'>⇢# Configuration Strategy</a></li> +<li>⇢ <a href='#-tempo-deployment-files'>⇢# Tempo Deployment Files</a></li> +<li>⇢ <a href='#-installation'>⇢# Installation</a></li> +<li>⇢ ⇢ <a href='#configuring-grafana-alloy-for-trace-collection'>Configuring Grafana Alloy for Trace Collection</a></li> +<li>⇢ <a href='#-otlp-receiver-configuration'>⇢# OTLP Receiver Configuration</a></li> +<li>⇢ <a href='#-upgrade-alloy'>⇢# Upgrade Alloy</a></li> +<li>⇢ ⇢ <a href='#demo-tracing-application'>Demo Tracing Application</a></li> +<li>⇢ <a href='#-application-architecture'>⇢# Application Architecture</a></li> +<li>⇢ ⇢ <a href='#visualizing-traces-in-grafana'>Visualizing Traces in Grafana</a></li> +<li>⇢ <a href='#-accessing-traces'>⇢# Accessing Traces</a></li> +<li>⇢ <a href='#-service-graph-visualization'>⇢# Service Graph Visualization</a></li> +<li>⇢ ⇢ <a href='#correlation-between-observability-signals'>Correlation Between Observability Signals</a></li> +<li>⇢ <a href='#-traces-to-logs'>⇢# Traces-to-Logs</a></li> +<li>⇢ <a href='#-traces-to-metrics'>⇢# Traces-to-Metrics</a></li> +<li>⇢ <a href='#-logs-to-traces'>⇢# Logs-to-Traces</a></li> +<li>⇢ ⇢ <a href='#generating-traces-for-testing'>Generating Traces for Testing</a></li> +<li>⇢ ⇢ <a href='#verifying-the-complete-pipeline'>Verifying the Complete Pipeline</a></li> +<li>⇢ ⇢ <a href='#practical-example-viewing-a-distributed-trace'>Practical Example: Viewing a Distributed Trace</a></li> +<li>⇢ ⇢ <a href='#storage-and-retention'>Storage and Retention</a></li> +<li>⇢ ⇢ <a href='#configuration-files'>Configuration Files</a></li> <li>⇢ <a href='#summary'>Summary</a></li> </ul><br /> <h2 style='display: inline' id='introduction'>Introduction</h2><br /> <br /> -<span>In this blog post, I set up a complete observability stack for the k3s cluster. Observability is crucial for understanding what's happening inside the cluster—whether its tracking resource usage, debugging issues, or analysing application behaviour. The stack consists of four main components, all deployed into the <span class='inlinecode'>monitoring</span> namespace:</span><br /> +<span>In this blog post, I set up a complete observability stack for the k3s cluster. Observability is crucial for understanding what's happening inside the cluster—whether its tracking resource usage, debugging issues, or analysing application behaviour. The stack consists of five main components, all deployed into the <span class='inlinecode'>monitoring</span> namespace:</span><br /> <br /> <ul> <li>Prometheus: time-series database for metrics collection and alerting</li> <li>Grafana: visualisation and dashboarding frontend</li> <li>Loki: log aggregation system (like Prometheus, but for logs)</li> -<li>Alloy: telemetry collector that ships logs from all pods to Loki</li> +<li>Alloy: telemetry collector that ships logs and traces from all pods to Loki and Tempo</li> +<li>Tempo: distributed tracing backend for request flow analysis across microservices</li> </ul><br /> -<span>Together, these form the "PLG" stack (Prometheus, Loki, Grafana), which is a popular open-source alternative to commercial observability platforms.</span><br /> +<span>Together, these form the "PLG" stack (Prometheus, Loki, Grafana) extended with Tempo for distributed tracing, which is a popular open-source alternative to commercial observability platforms.</span><br /> <br /> <span>All manifests for the f3s stack live in my configuration repository:</span><br /> <br /> @@ -120,6 +153,7 @@ http://www.gnu.org/software/src-highlite --> <li><span class='inlinecode'>/data/nfs/k3svolumes/prometheus/data</span> — Prometheus time-series database</li> <li><span class='inlinecode'>/data/nfs/k3svolumes/grafana/data</span> — Grafana configuration, dashboards, and plugins</li> <li><span class='inlinecode'>/data/nfs/k3svolumes/loki/data</span> — Loki log chunks and index</li> +<li><span class='inlinecode'>/data/nfs/k3svolumes/tempo/data</span> — Tempo trace data and WAL</li> </ul><br /> <span>Each path gets a corresponding <span class='inlinecode'>PersistentVolume</span> and <span class='inlinecode'>PersistentVolumeClaim</span> in Kubernetes, allowing pods to mount them as regular volumes. Because the underlying storage is ZFS with replication, we get snapshots and redundancy for free.</span><br /> <br /> @@ -218,7 +252,7 @@ kubeControllerManager: insecureSkipVerify: true </pre> <br /> -<span>By default, k3s binds the controller-manager to localhost only, so the "Kubernetes / Controller Manager" dashboard in Grafana will show no data. To expose the metrics endpoint, add the following to <span class='inlinecode'>/etc/rancher/k3s/config.yaml</span> on each k3s server node:</span><br /> +<span>By default, k3s binds the controller-manager to localhost only and doesn't expose etcd metrics, so the "Kubernetes / Controller Manager" and "etcd" dashboards in Grafana will show no data. To fix both, add the following to <span class='inlinecode'>/etc/rancher/k3s/config.yaml</span> on each k3s server node:</span><br /> <br /> <!-- Generator: GNU source-highlight 3.1.9 by Lorenzo Bettini @@ -227,11 +261,26 @@ http://www.gnu.org/software/src-highlite --> <pre><font color="#F3E651">[</font><font color="#ff0000">root@r0 </font><font color="#F3E651">~]</font><i><font color="#ababab"># cat >> /etc/rancher/k3s/config.yaml << 'EOF'</font></i> <font color="#ff0000">kube-controller-manager-arg</font><font color="#F3E651">:</font> <font color="#ff0000"> - bind-address</font><font color="#F3E651">=</font><font color="#bb00ff">0.0</font><font color="#F3E651">.</font><font color="#bb00ff">0.0</font> +<font color="#ff0000">etcd-expose-metrics</font><font color="#F3E651">:</font><font color="#ff0000"> </font><b><font color="#ffffff">true</font></b> <font color="#ff0000">EOF</font> <font color="#F3E651">[</font><font color="#ff0000">root@r0 </font><font color="#F3E651">~]</font><i><font color="#ababab"># systemctl restart k3s</font></i> </pre> <br /> -<span>Repeat for <span class='inlinecode'>r1</span> and <span class='inlinecode'>r2</span>. After restarting all nodes, the controller-manager metrics endpoint will be accessible and Prometheus can scrape it.</span><br /> +<span>Repeat for <span class='inlinecode'>r1</span> and <span class='inlinecode'>r2</span>. After restarting all nodes, the controller-manager metrics endpoint will be accessible and etcd metrics are available on port 2381. Prometheus can now scrape both.</span><br /> +<br /> +<span>Verify etcd metrics are exposed:</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><font color="#F3E651">[</font><font color="#ff0000">root@r0 </font><font color="#F3E651">~]</font><i><font color="#ababab"># curl -s http://127.0.0.1:2381/metrics | grep etcd_server_has_leader</font></i> +<font color="#ff0000">etcd_server_has_leader </font><font color="#bb00ff">1</font> +</pre> +<br /> +<span>The full <span class='inlinecode'>persistence-values.yaml</span> and all other Prometheus configuration files are available on Codeberg:</span><br /> +<br /> +<a class='textlink' href='https://codeberg.org/snonux/conf/src/branch/master/f3s/prometheus'>codeberg.org/snonux/conf/f3s/prometheus</a><br /> <br /> <span>The persistent volume definitions bind to specific paths on the NFS share using <span class='inlinecode'>hostPath</span> volumes—the same pattern used for other services in Part 7:</span><br /> <br /> @@ -258,6 +307,8 @@ http://www.gnu.org/software/src-highlite --> <br /> <a href='./f3s-kubernetes-with-freebsd-part-8/grafana-dashboard.png'><img alt='Grafana dashboard showing cluster metrics' title='Grafana dashboard showing cluster metrics' src='./f3s-kubernetes-with-freebsd-part-8/grafana-dashboard.png' /></a><br /> <br /> +<a href='./f3s-kubernetes-with-freebsd-part-8/grafana-etcd-dashboard.png'><img alt='Grafana etcd dashboard showing cluster health, RPC rate, disk sync duration, and peer round trip times' title='Grafana etcd dashboard showing cluster health, RPC rate, disk sync duration, and peer round trip times' src='./f3s-kubernetes-with-freebsd-part-8/grafana-etcd-dashboard.png' /></a><br /> +<br /> <h2 style='display: inline' id='installing-loki-and-alloy'>Installing Loki and Alloy</h2><br /> <br /> <span>While Prometheus handles metrics, Loki handles logs. It's designed to be cost-effective and easy to operate—it doesn't index the contents of logs, only the metadata (labels), making it very efficient for storage.</span><br /> @@ -409,8 +460,11 @@ http://www.gnu.org/software/src-highlite --> <font color="#ff0000">prometheus-prometheus-node-exporter-2nsg9 </font><font color="#bb00ff">1</font><font color="#F3E651">/</font><font color="#bb00ff">1</font><font color="#ff0000"> Running </font><font color="#bb00ff">0</font><font color="#ff0000"> 42d</font> <font color="#ff0000">prometheus-prometheus-node-exporter-mqr</font><font color="#bb00ff">25</font><font color="#ff0000"> </font><font color="#bb00ff">1</font><font color="#F3E651">/</font><font color="#bb00ff">1</font><font color="#ff0000"> Running </font><font color="#bb00ff">0</font><font color="#ff0000"> 42d</font> <font color="#ff0000">prometheus-prometheus-node-exporter-wp4ds </font><font color="#bb00ff">1</font><font color="#F3E651">/</font><font color="#bb00ff">1</font><font color="#ff0000"> Running </font><font color="#bb00ff">0</font><font color="#ff0000"> 42d</font> +<font color="#ff0000">tempo-</font><font color="#bb00ff">0</font><font color="#ff0000"> </font><font color="#bb00ff">1</font><font color="#F3E651">/</font><font color="#bb00ff">1</font><font color="#ff0000"> Running </font><font color="#bb00ff">0</font><font color="#ff0000"> 1d</font> </pre> <br /> +<span>Note: Tempo (<span class='inlinecode'>tempo-0</span>) is deployed later in this post in the "Distributed Tracing with Grafana Tempo" section. It is included in the pod listing here for completeness.</span><br /> +<br /> <span>And the services:</span><br /> <br /> <!-- Generator: GNU source-highlight 3.1.9 @@ -429,6 +483,7 @@ http://www.gnu.org/software/src-highlite --> <font color="#ff0000">prometheus-kube-prometheus-prometheus ClusterIP </font><font color="#bb00ff">10.43</font><font color="#F3E651">.</font><font color="#bb00ff">152.163</font><font color="#ff0000"> </font><font color="#bb00ff">9090</font><font color="#ff0000">/TCP</font><font color="#F3E651">,</font><font color="#bb00ff">8080</font><font color="#ff0000">/TCP</font> <font color="#ff0000">prometheus-kube-state-metrics ClusterIP </font><font color="#bb00ff">10.43</font><font color="#F3E651">.</font><font color="#bb00ff">64.26</font><font color="#ff0000"> </font><font color="#bb00ff">8080</font><font color="#ff0000">/TCP</font> <font color="#ff0000">prometheus-prometheus-node-exporter ClusterIP </font><font color="#bb00ff">10.43</font><font color="#F3E651">.</font><font color="#bb00ff">127.242</font><font color="#ff0000"> </font><font color="#bb00ff">9100</font><font color="#ff0000">/TCP</font> +<font color="#ff0000">tempo ClusterIP </font><font color="#bb00ff">10.43</font><font color="#F3E651">.</font><font color="#bb00ff">91.44</font><font color="#ff0000"> </font><font color="#bb00ff">3200</font><font color="#ff0000">/TCP</font><font color="#F3E651">,</font><font color="#bb00ff">4317</font><font color="#ff0000">/TCP</font><font color="#F3E651">,</font><font color="#bb00ff">4318</font><font color="#ff0000">/TCP</font> </pre> <br /> <span>Let me break down what each pod does:</span><br /> @@ -457,6 +512,9 @@ http://www.gnu.org/software/src-highlite --> <ul> <li><span class='inlinecode'>prometheus-prometheus-node-exporter-...</span>: three Node Exporter pods running as a DaemonSet, one on each node. They expose hardware and OS-level metrics: CPU usage, memory, disk I/O, filesystem usage, network statistics, and more. These feed the "Node Exporter" dashboards in Grafana.</li> </ul><br /> +<ul> +<li><span class='inlinecode'>tempo-0</span>: the Grafana Tempo instance for distributed tracing. It receives trace data from Alloy via OTLP (OpenTelemetry Protocol), stores traces on the NFS-backed persistent volume, and serves queries to Grafana. Tempo is covered in detail in the "Distributed Tracing with Grafana Tempo" section later in this post.</li> +</ul><br /> <h2 style='display: inline' id='using-the-observability-stack'>Using the observability stack</h2><br /> <br /> <h3 style='display: inline' id='viewing-metrics-in-grafana'>Viewing metrics in Grafana</h3><br /> @@ -642,7 +700,313 @@ spec: <br /> <span>Unlike memory metrics, disk I/O metrics (<span class='inlinecode'>node_disk_read_bytes_total</span>, <span class='inlinecode'>node_disk_written_bytes_total</span>, etc.) are not available on FreeBSD. The Linux diskstats collector that provides these metrics doesn't have a FreeBSD equivalent in the node_exporter.</span><br /> <br /> -<span>The disk I/O panels in the Node Exporter dashboards will show "No data" for FreeBSD hosts. FreeBSD does expose ZFS-specific metrics (<span class='inlinecode'>node_zfs_arcstats_*</span>) for ARC cache performance, and per-dataset I/O stats are available via <span class='inlinecode'>sysctl kstat.zfs</span>, but mapping these to the Linux-style metrics the dashboards expect is non-trivial. Creating custom ZFS-specific dashboards is left as an exercise for another day.</span><br /> +<span>The disk I/O panels in the Node Exporter dashboards will show "No data" for FreeBSD hosts. FreeBSD does expose ZFS-specific metrics (<span class='inlinecode'>node_zfs_arcstats_*</span>) for ARC cache performance, and per-dataset I/O stats are available via <span class='inlinecode'>sysctl kstat.zfs</span>, but mapping these to the Linux-style metrics the dashboards expect is non-trivial. To address this, I created custom ZFS-specific dashboards, covered in the next section.</span><br /> +<br /> +<h2 style='display: inline' id='zfs-monitoring-for-freebsd-servers'>ZFS Monitoring for FreeBSD Servers</h2><br /> +<br /> +<span>The FreeBSD servers (f0, f1, f2) that provide NFS storage to the k3s cluster have ZFS filesystems. Monitoring ZFS performance is crucial for understanding storage performance and cache efficiency.</span><br /> +<br /> +<h3 style='display: inline' id='node-exporter-zfs-collector'>Node Exporter ZFS Collector</h3><br /> +<br /> +<span>The node_exporter running on each FreeBSD server (v1.9.1) includes a built-in ZFS collector that exposes metrics via sysctls. The ZFS collector is enabled by default and provides:</span><br /> +<br /> +<ul> +<li>ARC (Adaptive Replacement Cache) statistics</li> +<li>Cache hit/miss rates</li> +<li>Memory usage and allocation</li> +<li>MRU/MFU cache breakdown</li> +<li>Data vs metadata distribution</li> +</ul><br /> +<h3 style='display: inline' id='verifying-zfs-metrics'>Verifying ZFS Metrics</h3><br /> +<br /> +<span>On any FreeBSD server, check that ZFS metrics are being exposed:</span><br /> +<br /> +<pre> +paul@f0:~ % curl -s http://localhost:9100/metrics | grep node_zfs_arcstats | wc -l + 69 +</pre> +<br /> +<span>The metrics are automatically scraped by Prometheus through the existing static configuration in additional-scrape-configs.yaml which targets all FreeBSD servers on port 9100 with the os: freebsd label.</span><br /> +<br /> +<h3 style='display: inline' id='zfs-recording-rules'>ZFS Recording Rules</h3><br /> +<br /> +<span>Created recording rules for easier dashboard consumption in zfs-recording-rules.yaml:</span><br /> +<br /> +<pre> +apiVersion: monitoring.coreos.com/v1 +kind: PrometheusRule +metadata: + name: freebsd-zfs-rules + namespace: monitoring + labels: + release: prometheus +spec: + groups: + - name: freebsd-zfs-arc + interval: 30s + rules: + - record: node_zfs_arc_hit_rate_percent + expr: | + 100 * ( + rate(node_zfs_arcstats_hits_total{os="freebsd"}[5m]) / + (rate(node_zfs_arcstats_hits_total{os="freebsd"}[5m]) + + rate(node_zfs_arcstats_misses_total{os="freebsd"}[5m])) + ) + labels: + os: freebsd + - record: node_zfs_arc_memory_usage_percent + expr: | + 100 * ( + node_zfs_arcstats_size_bytes{os="freebsd"} / + node_zfs_arcstats_c_max_bytes{os="freebsd"} + ) + labels: + os: freebsd + # Additional rules for metadata %, target %, MRU/MFU %, etc. +</pre> +<br /> +<span>These recording rules calculate:</span><br /> +<br /> +<ul> +<li>ARC hit rate percentage</li> +<li>ARC memory usage percentage (current vs maximum)</li> +<li>ARC target percentage (target vs maximum)</li> +<li>Metadata vs data percentages</li> +<li>MRU vs MFU cache percentages</li> +<li>Demand data and metadata hit rates</li> +</ul><br /> +<h3 style='display: inline' id='grafana-dashboards'>Grafana Dashboards</h3><br /> +<br /> +<span>Created two comprehensive ZFS monitoring dashboards (zfs-dashboards.yaml):</span><br /> +<br /> +<span>**Dashboard 1: FreeBSD ZFS (per-host detailed view)**</span><br /> +<br /> +<span>Includes variables to select:</span><br /> +<br /> +<ul> +<li>FreeBSD server (f0, f1, or f2)</li> +<li>ZFS pool (zdata, zroot, or all)</li> +</ul><br /> +<span>Pool Overview Row:</span><br /> +<br /> +<ul> +<li>Pool Capacity gauge (with thresholds: green <70%, yellow <85%, red >85%)</li> +<li>Pool Health status (ONLINE/DEGRADED/FAULTED with color coding)</li> +<li>Total Pool Size stat</li> +<li>Free Space stat</li> +<li>Pool Space Usage Over Time (stacked: used + free)</li> +<li>Pool Capacity Trend time series</li> +</ul><br /> +<span>Dataset Statistics Row:</span><br /> +<br /> +<ul> +<li>Table showing all datasets with columns: Pool, Dataset, Used, Available, Referenced</li> +<li>Automatically filters by selected pool</li> +</ul><br /> +<span>ARC Cache Statistics Row:</span><br /> +<br /> +<ul> +<li>ARC Hit Rate gauge (red <70%, yellow <90%, green >=90%)</li> +<li>ARC Size time series (current, target, max)</li> +<li>ARC Memory Usage percentage gauge</li> +<li>ARC Hits vs Misses rate</li> +<li>ARC Data vs Metadata stacked time series</li> +</ul><br /> +<span>**Dashboard 2: FreeBSD ZFS Summary (cluster-wide overview)**</span><br /> +<br /> +<span>Cluster-Wide Pool Statistics Row:</span><br /> +<br /> +<ul> +<li>Total Storage Capacity across all servers</li> +<li>Total Used space</li> +<li>Total Free space</li> +<li>Average Pool Capacity gauge</li> +<li>Pool Health Status (worst case across cluster)</li> +<li>Total Pool Space Usage Over Time</li> +<li>Per-Pool Capacity time series (all pools on all hosts)</li> +</ul><br /> +<span>Per-Host Pool Breakdown Row:</span><br /> +<br /> +<ul> +<li>Bar gauge showing capacity by host and pool</li> +<li>Table with all pools: Host, Pool, Size, Used, Free, Capacity %, Health</li> +</ul><br /> +<span>Cluster-Wide ARC Statistics Row:</span><br /> +<br /> +<ul> +<li>Average ARC Hit Rate gauge across all hosts</li> +<li>ARC Hit Rate by Host time series</li> +<li>Total ARC Size Across Cluster</li> +<li>Total ARC Hits vs Misses (cluster-wide sum)</li> +<li>ARC Size by Host</li> +</ul><br /> +<span>Dashboard Visualization:</span><br /> +<br /> +<a href='./f3s-kubernetes-with-freebsd-part-8/grafana-zfs-dashboard.png'><img alt='ZFS monitoring dashboard in Grafana showing pool capacity, health, and I/O throughput' title='ZFS monitoring dashboard in Grafana showing pool capacity, health, and I/O throughput' src='./f3s-kubernetes-with-freebsd-part-8/grafana-zfs-dashboard.png' /></a><br /> +<a href='./f3s-kubernetes-with-freebsd-part-8/grafana-zfs-arc-stats.png'><img alt='ZFS ARC cache statistics showing hit rate, memory usage, and size trends' title='ZFS ARC cache statistics showing hit rate, memory usage, and size trends' src='./f3s-kubernetes-with-freebsd-part-8/grafana-zfs-arc-stats.png' /></a><br /> +<a href='./f3s-kubernetes-with-freebsd-part-8/grafana-zfs-datasets.png'><img alt='ZFS datasets table and ARC data vs metadata breakdown' title='ZFS datasets table and ARC data vs metadata breakdown' src='./f3s-kubernetes-with-freebsd-part-8/grafana-zfs-datasets.png' /></a><br /> +<br /> +<h3 style='display: inline' id='deployment'>Deployment</h3><br /> +<br /> +<span>Applied the resources to the cluster:</span><br /> +<br /> +<pre> +cd /home/paul/git/conf/f3s/prometheus +kubectl apply -f zfs-recording-rules.yaml +kubectl apply -f zfs-dashboards.yaml +</pre> +<br /> +<span>Updated Justfile to include ZFS recording rules in install and upgrade targets:</span><br /> +<br /> +<pre> +install: + kubectl apply -f persistent-volumes.yaml + kubectl create secret generic additional-scrape-configs --from-file=additional-scrape-configs.yaml -n monitoring --dry-run=client -o yaml | kubectl apply -f - + helm install prometheus prometheus-community/kube-prometheus-stack --namespace monitoring -f persistence-values.yaml + kubectl apply -f freebsd-recording-rules.yaml + kubectl apply -f openbsd-recording-rules.yaml + kubectl apply -f zfs-recording-rules.yaml + just -f grafana-ingress/Justfile install +</pre> +<br /> +<h3 style='display: inline' id='verifying-zfs-metrics-in-prometheus'>Verifying ZFS Metrics in Prometheus</h3><br /> +<br /> +<span>Check that ZFS metrics are being collected:</span><br /> +<br /> +<pre> +kubectl exec -n monitoring prometheus-prometheus-kube-prometheus-prometheus-0 -c prometheus -- \ + wget -qO- 'http://localhost:9090/api/v1/query?query=node_zfs_arcstats_size_bytes' +</pre> +<br /> +<span>Check recording rules are calculating correctly:</span><br /> +<br /> +<pre> +kubectl exec -n monitoring prometheus-prometheus-kube-prometheus-prometheus-0 -c prometheus -- \ + wget -qO- 'http://localhost:9090/api/v1/query?query=node_zfs_arc_memory_usage_percent' +</pre> +<br /> +<span>Example output shows memory usage percentage for each FreeBSD server:</span><br /> +<br /> +<pre> +"result":[ + {"metric":{"instance":"192.168.2.130:9100","os":"freebsd"},"value":[...,"37.58"]}, + {"metric":{"instance":"192.168.2.131:9100","os":"freebsd"},"value":[...,"12.85"]}, + {"metric":{"instance":"192.168.2.132:9100","os":"freebsd"},"value":[...,"13.44"]} +] +</pre> +<br /> +<h3 style='display: inline' id='key-metrics-to-monitor'>Key Metrics to Monitor</h3><br /> +<br /> +<ul> +<li>ARC Hit Rate: Should typically be above 90% for optimal performance. Lower hit rates indicate the ARC cache is too small or workload has poor locality.</li> +<li>ARC Memory Usage: Shows how much of the maximum ARC size is being used. If consistently at or near maximum, the ARC is effectively utilizing available memory.</li> +<li>Data vs Metadata: Typically data should dominate, but workloads with many small files will show higher metadata percentages.</li> +<li>MRU vs MFU: Most Recently Used vs Most Frequently Used cache. The ratio depends on workload characteristics.</li> +<li>Pool Capacity: Monitor pool usage to ensure adequate free space. ZFS performance degrades when pools exceed 80% capacity.</li> +<li>Pool Health: Should always show ONLINE (green). DEGRADED (yellow) indicates a disk issue requiring attention. FAULTED (red) requires immediate action.</li> +<li>Dataset Usage: Track which datasets are consuming the most space to identify growth trends and plan capacity.</li> +</ul><br /> +<h3 style='display: inline' id='zfs-pool-and-dataset-metrics-via-textfile-collector'>ZFS Pool and Dataset Metrics via Textfile Collector</h3><br /> +<br /> +<span>To complement the ARC statistics from node_exporter's built-in ZFS collector, I added pool capacity and dataset metrics using the textfile collector feature.</span><br /> +<br /> +<span>Created a script at <span class='inlinecode'>/usr/local/bin/zfs_pool_metrics.sh</span> on each FreeBSD server:</span><br /> +<br /> +<pre> +#!/bin/sh +# ZFS Pool and Dataset Metrics Collector for Prometheus + +OUTPUT_FILE="/var/tmp/node_exporter/zfs_pools.prom.$$" +FINAL_FILE="/var/tmp/node_exporter/zfs_pools.prom" + +mkdir -p /var/tmp/node_exporter + +{ + # Pool metrics + echo "# HELP zfs_pool_size_bytes Total size of ZFS pool" + echo "# TYPE zfs_pool_size_bytes gauge" + echo "# HELP zfs_pool_allocated_bytes Allocated space in ZFS pool" + echo "# TYPE zfs_pool_allocated_bytes gauge" + echo "# HELP zfs_pool_free_bytes Free space in ZFS pool" + echo "# TYPE zfs_pool_free_bytes gauge" + echo "# HELP zfs_pool_capacity_percent Capacity percentage" + echo "# TYPE zfs_pool_capacity_percent gauge" + echo "# HELP zfs_pool_health Pool health (0=ONLINE, 1=DEGRADED, 2=FAULTED)" + echo "# TYPE zfs_pool_health gauge" + + zpool list -Hp -o name,size,allocated,free,capacity,health | \ + while IFS=$'\t' read name size alloc free cap health; do + case "$health" in + ONLINE) health_val=0 ;; + DEGRADED) health_val=1 ;; + FAULTED) health_val=2 ;; + *) health_val=6 ;; + esac + cap_num=$(echo "$cap" | sed 's/%//') + + echo "zfs_pool_size_bytes{pool=\"$name\"} $size" + echo "zfs_pool_allocated_bytes{pool=\"$name\"} $alloc" + echo "zfs_pool_free_bytes{pool=\"$name\"} $free" + echo "zfs_pool_capacity_percent{pool=\"$name\"} $cap_num" + echo "zfs_pool_health{pool=\"$name\"} $health_val" + done + + # Dataset metrics + echo "# HELP zfs_dataset_used_bytes Used space in dataset" + echo "# TYPE zfs_dataset_used_bytes gauge" + echo "# HELP zfs_dataset_available_bytes Available space" + echo "# TYPE zfs_dataset_available_bytes gauge" + echo "# HELP zfs_dataset_referenced_bytes Referenced space" + echo "# TYPE zfs_dataset_referenced_bytes gauge" + + zfs list -Hp -t filesystem -o name,used,available,referenced | \ + while IFS=$'\t' read name used avail ref; do + pool=$(echo "$name" | cut -d/ -f1) + echo "zfs_dataset_used_bytes{pool=\"$pool\",dataset=\"$name\"} $used" + echo "zfs_dataset_available_bytes{pool=\"$pool\",dataset=\"$name\"} $avail" + echo "zfs_dataset_referenced_bytes{pool=\"$pool\",dataset=\"$name\"} $ref" + done +} > "$OUTPUT_FILE" + +mv "$OUTPUT_FILE" "$FINAL_FILE" +</pre> +<br /> +<span>Deployed to all FreeBSD servers:</span><br /> +<br /> +<pre> +for host in f0 f1 f2; do + scp /tmp/zfs_pool_metrics.sh paul@$host:/tmp/ + ssh paul@$host 'doas mv /tmp/zfs_pool_metrics.sh /usr/local/bin/ && \ + doas chmod +x /usr/local/bin/zfs_pool_metrics.sh' +done +</pre> +<br /> +<span>Set up cron jobs to run every minute:</span><br /> +<br /> +<pre> +for host in f0 f1 f2; do + ssh paul@$host 'echo "* * * * * /usr/local/bin/zfs_pool_metrics.sh >/dev/null 2>&1" | \ + doas crontab -' +done +</pre> +<br /> +<span>The textfile collector (already configured with --collector.textfile.directory=/var/tmp/node_exporter) automatically picks up the metrics.</span><br /> +<br /> +<span>Verify metrics are being exposed:</span><br /> +<br /> +<pre> +paul@f0:~ % curl -s http://localhost:9100/metrics | grep "^zfs_pool" | head -5 +zfs_pool_allocated_bytes{pool="zdata"} 6.47622733824e+11 +zfs_pool_allocated_bytes{pool="zroot"} 5.3338578944e+10 +zfs_pool_capacity_percent{pool="zdata"} 64 +zfs_pool_capacity_percent{pool="zroot"} 10 +zfs_pool_free_bytes{pool="zdata"} 3.48809678848e+11 +</pre> +<br /> +<span>All ZFS-related configuration files are available on Codeberg:</span><br /> +<br /> +<a class='textlink' href='https://codeberg.org/snonux/conf/src/branch/master/f3s/prometheus/zfs-recording-rules.yaml'>zfs-recording-rules.yaml on Codeberg</a><br /> +<a class='textlink' href='https://codeberg.org/snonux/conf/src/branch/master/f3s/prometheus/zfs-dashboards.yaml'>zfs-dashboards.yaml on Codeberg</a><br /> <br /> <h2 style='display: inline' id='monitoring-external-openbsd-hosts'>Monitoring external OpenBSD hosts</h2><br /> <br /> @@ -769,18 +1133,671 @@ spec: <br /> <span>After running <span class='inlinecode'>just upgrade</span>, the OpenBSD hosts appear in Prometheus targets and the Node Exporter dashboards.</span><br /> <br /> +<h2 style='display: inline' id='distributed-tracing-with-grafana-tempo'>Distributed Tracing with Grafana Tempo</h2><br /> +<br /> +<span>After implementing logs (Loki) and metrics (Prometheus), the final pillar of observability is distributed tracing. Grafana Tempo provides distributed tracing capabilities that help understand request flows across microservices.</span><br /> +<br /> +<span>For a preview of what distributed tracing with Tempo looks like in Grafana, see the X-RAG blog post:</span><br /> +<br /> +<a class='textlink' href='./2025-12-24-x-rag-observability-hackathon.html'>X-RAG Observability Hackathon</a><br /> +<br /> +<h3 style='display: inline' id='why-distributed-tracing'>Why Distributed Tracing?</h3><br /> +<br /> +<span>In a microservices architecture, a single user request may traverse multiple services. Distributed tracing:</span><br /> +<br /> +<ul> +<li>Tracks requests across service boundaries</li> +<li>Identifies performance bottlenecks</li> +<li>Visualizes service dependencies</li> +<li>Correlates with logs and metrics</li> +<li>Helps debug complex distributed systems</li> +</ul><br /> +<h3 style='display: inline' id='deploying-grafana-tempo'>Deploying Grafana Tempo</h3><br /> +<br /> +<span>Tempo is deployed in monolithic mode, following the same pattern as Loki's SingleBinary deployment.</span><br /> +<br /> +<span>#### Configuration Strategy</span><br /> +<br /> +<span>**Deployment Mode:** Monolithic (all components in one process)</span><br /> +<ul> +<li>Simpler operation than microservices mode</li> +<li>Suitable for the cluster scale</li> +<li>Consistent with Loki deployment pattern</li> +</ul><br /> +<span>**Storage:** Filesystem backend using hostPath</span><br /> +<ul> +<li>10Gi storage at /data/nfs/k3svolumes/tempo/data</li> +<li>7-day retention (168h)</li> +<li>Local storage is the only option for monolithic mode</li> +</ul><br /> +<span>**OTLP Receivers:** Standard OpenTelemetry Protocol ports</span><br /> +<ul> +<li>gRPC: 4317</li> +<li>HTTP: 4318</li> +<li>Bind to 0.0.0.0 to avoid Tempo 2.7+ localhost-only binding issue</li> +</ul><br /> +<span>#### Tempo Deployment Files</span><br /> +<br /> +<span>Created in /home/paul/git/conf/f3s/tempo/:</span><br /> +<br /> +<span>**values.yaml** - Helm chart configuration:</span><br /> +<br /> +<pre> +tempo: + retention: 168h + storage: + trace: + backend: local + local: + path: /var/tempo/traces + wal: + path: /var/tempo/wal + receivers: + otlp: + protocols: + grpc: + endpoint: 0.0.0.0:4317 + http: + endpoint: 0.0.0.0:4318 + +persistence: + enabled: true + size: 10Gi + storageClassName: "" + +resources: + limits: + cpu: 1000m + memory: 2Gi + requests: + cpu: 500m + memory: 1Gi +</pre> +<br /> +<span>**persistent-volumes.yaml** - Storage configuration:</span><br /> +<br /> +<pre> +apiVersion: v1 +kind: PersistentVolume +metadata: + name: tempo-data-pv +spec: + capacity: + storage: 10Gi + accessModes: + - ReadWriteOnce + persistentVolumeReclaimPolicy: Retain + hostPath: + path: /data/nfs/k3svolumes/tempo/data +--- +apiVersion: v1 +kind: PersistentVolumeClaim +metadata: + name: tempo-data-pvc + namespace: monitoring +spec: + storageClassName: "" + accessModes: + - ReadWriteOnce + resources: + requests: + storage: 10Gi +</pre> +<br /> +<span>**Grafana Datasource Provisioning**</span><br /> +<br /> +<span>All Grafana datasources (Prometheus, Alertmanager, Loki, Tempo) are provisioned via a unified ConfigMap that is directly mounted to the Grafana pod. This approach ensures datasources are loaded on startup without requiring sidecar-based discovery.</span><br /> +<br /> +<span>In /home/paul/git/conf/f3s/prometheus/grafana-datasources-all.yaml:</span><br /> +<br /> +<pre> +apiVersion: v1 +kind: ConfigMap +metadata: + name: grafana-datasources-all + namespace: monitoring +data: + datasources.yaml: | + apiVersion: 1 + datasources: + - name: Prometheus + type: prometheus + uid: prometheus + url: http://prometheus-kube-prometheus-prometheus.monitoring:9090/ + access: proxy + isDefault: true + - name: Alertmanager + type: alertmanager + uid: alertmanager + url: http://prometheus-kube-prometheus-alertmanager.monitoring:9093/ + - name: Loki + type: loki + uid: loki + url: http://loki.monitoring.svc.cluster.local:3100 + - name: Tempo + type: tempo + uid: tempo + url: http://tempo.monitoring.svc.cluster.local:3200 + jsonData: + tracesToLogsV2: + datasourceUid: loki + spanStartTimeShift: -1h + spanEndTimeShift: 1h + tracesToMetrics: + datasourceUid: prometheus + serviceMap: + datasourceUid: prometheus + nodeGraph: + enabled: true +</pre> +<br /> +<span>The kube-prometheus-stack Helm values (persistence-values.yaml) are configured to:</span><br /> +<ul> +<li>Disable sidecar-based datasource provisioning</li> +<li>Mount grafana-datasources-all ConfigMap directly to /etc/grafana/provisioning/datasources/</li> +</ul><br /> +<span>This direct mounting approach is simpler and more reliable than sidecar-based discovery.</span><br /> +<br /> +<span>#### Installation</span><br /> +<br /> +<pre> +cd /home/paul/git/conf/f3s/tempo +just install +</pre> +<br /> +<span>Verify Tempo is running:</span><br /> +<br /> +<pre> +kubectl get pods -n monitoring -l app.kubernetes.io/name=tempo +kubectl exec -n monitoring <tempo-pod> -- wget -qO- http://localhost:3200/ready +</pre> +<br /> +<h3 style='display: inline' id='configuring-grafana-alloy-for-trace-collection'>Configuring Grafana Alloy |