github.com/NVIDIA/aistore@v1.3.23-0.20240517131212-7df6609be51d/docs/prometheus.md (about) 1 --- 2 layout: post 3 title: PROMETHEUS 4 permalink: /docs/prometheus 5 redirect_from: 6 - /prometheus.md/ 7 - /docs/prometheus.md/ 8 --- 9 10 ## Monitoring AIStore with Prometheus 11 12 AIStore tracks a growing list of performance counters, utilization percentages, latency and throughput metrics, transmitted and received stats (total bytes and numbers of objects), error counters, and more. 13 14 Viewership is equally supported via: 15 * AIS node logs 16 * [CLI](/docs/cli.md), and specifically 17 * [`ais show cluster stats`](/docs/cli/cluster.md) command 18 19 On the monitoring backend side, AIS equally supports: 20 * [StatsD](https://github.com/etsy/statsd) with any compliant backend (e.g., Graphite/Grafana), and 21 * [Prometheus](https://prometheus.io/) 22 23 This document mostly talks about the "Prometheus" option. Other related documentation includes [AIS metrics](metrics.md) readme that provides general background, naming conventions and examples, and also have a separate section on `aisloader` metrics - the metrics generated by `aisloader` when running its benches. 24 25 > For `aisloader`, please refer to [Load Generator](/docs/aisloader.md) and [How To Benchmark AIStore](howto_benchmark.md). 26 27 ## Prometheus Exporter 28 29 AIStore is a fully compliant [Prometheus exporter](https://prometheus.io/docs/instrumenting/writing_exporters/) that natively supports [Prometheus](https://prometheus.io/) stats collection. There's no special configuration - the only thing required to enable the corresponding integration is letting AIStore know whether to publish its stats via StatsD **or** Prometheus. 30 31 The corresponding binary choice between StatsD and Prometheus is a **deployment-time** switch that is a single environment variable: **AIS_PROMETHEUS**. When a starting-up AIS node (gateway or storage target) sees `AIS_PROMETHEUS` in the environment it registers all its metric descriptions (names, labels, and helps) with Prometheus and provides HTTP endpoint `/metrics` for subsequent collection (aka "scraping") by Prometheus. 32 33 > With no `AIS_PROMETHEUS` in the environment, AIS nodes default to StatsD. 34 35 Here's a simplified example: 36 37 ```console 38 $ AIS_PROMETHEUS=true aisnode -config=/etc/ais/ais.json -local_config=/etc/ais/ais_local.json -role=target 39 40 # Assuming the target with hostname "hostname" listens on port 8081: 41 $ curl http://hostname:8081/metrics | grep ais 42 43 # A sample output follows below (note the metric names that must be self-explanatory): 44 45 # TYPE ais_target_DFIltrTgz_disk_sda_avg_rsize gauge 46 ais_target_DFIltrTgz_disk_sda_avg_rsize 23560 47 # HELP ais_target_DFIltrTgz_disk_sda_avg_wsize average write size (bytes) 48 # TYPE ais_target_DFIltrTgz_disk_sda_avg_wsize gauge 49 ais_target_DFIltrTgz_disk_sda_avg_wsize 63120 50 # HELP ais_target_DFIltrTgz_disk_sda_util gauge 51 # TYPE ais_target_DFIltrTgz_disk_sda_util gauge 52 ais_target_DFIltrTgz_disk_sda_util 42 53 # HELP ais_target_DFIltrTgz_get_mbps throughput (MB/s) 54 # TYPE ais_target_DFIltrTgz_get_mbps gauge 55 ais_target_DFIltrTgz_get_mbps 72.65 56 # HELP ais_target_DFIltrTgz_get_ms latency (milliseconds) 57 # TYPE ais_target_DFIltrTgz_get_ms gauge 58 ais_target_DFIltrTgz_get_ms 2 59 # HELP ais_target_DFIltrTgz_get_n total number of operations 60 # TYPE ais_target_DFIltrTgz_get_n counter 61 ais_target_DFIltrTgz_get_n 155431 62 # HELP ais_target_DFIltrTgz_get_redir_ms latency (milliseconds) 63 # TYPE ais_target_DFIltrTgz_get_redir_ms gauge 64 ais_target_DFIltrTgz_get_redir_ms 0 65 # HELP ais_target_DFIltrTgz_kalive_ms latency (milliseconds) 66 # TYPE ais_target_DFIltrTgz_kalive_ms gauge 67 ais_target_DFIltrTgz_kalive_ms 1 68 # HELP ais_target_DFIltrTgz_lst_ms latency (milliseconds) 69 # TYPE ais_target_DFIltrTgz_lst_ms gauge 70 ais_target_DFIltrTgz_lst_ms 2 71 # HELP ais_target_DFIltrTgz_lst_n total number of operations 72 # TYPE ais_target_DFIltrTgz_lst_n counter 73 ais_target_DFIltrTgz_lst_n 120 74 # HELP ais_target_DFIltrTgz_put_ms latency (milliseconds) 75 # TYPE ais_target_DFIltrTgz_put_ms gauge 76 ais_target_DFIltrTgz_put_ms 5 77 # HELP ais_target_DFIltrTgz_put_n total number of operations 78 ... 79 ``` 80 81 References: 82 83 * https://prometheus.io/docs/instrumenting/writing_exporters/ 84 * https://prometheus.io/docs/concepts/data_model/ 85 * https://prometheus.io/docs/concepts/metric_types/ 86 87 ## StatsD Exporter for Prometheus 88 89 If, for whatever reason, you decide to use the "StatsD" option, you can still send AIS stats to Prometheus - via its own generic [statsd_exporter](https://github.com/prometheus/statsd_exporter) extension that on-the-fly translates StatsD formatted metrics. 90 91 > **Note**: while native Prometheus integration (the previous section) is the preferred and recommended option [statsd_exporter](https://github.com/prometheus/statsd_exporter) can be considered a backup plan for deployments with very special requirements. 92 93 First, the picture: 94 95 ![AIStore monitoring with Prometheus](images/statsd-exporter.png) 96 97 The diagram depicts AIS cluster that runs an arbitrary number of nodes with each node periodically sending its StatsD metrics to a configured UDP address of any compliant StatsD server. In fact, [statsd_exporter](https://github.com/prometheus/statsd_exporter) is one such compliant StatsD server that happens to be available out of the box. 98 99 To deploy [statsd_exporter](https://github.com/prometheus/statsd_exporter): 100 101 * you could either use [prebuilt container image](https://quay.io/repository/prometheus/statsd-exporter); 102 * or, `git clone` or `go install` the exporter's own repository at https://github.com/prometheus/statsd_exporter and then run it as shown above. Just take a note of the default StatsD port: **8125**. 103 104 To test a combination of AIStore and [statsd_exporter](https://github.com/prometheus/statsd_exporter) without Prometheus, run the exporter with debug: 105 106 ```console 107 $ statsd_exporter --statsd.listen-udp localhost:8125 --log.level debug 108 ``` 109 110 The resulting (debug) output will look something like: 111 112 ```console 113 level=info ts=2021-05-13T15:30:22.251Z caller=main.go:321 msg="Starting StatsD -> Prometheus Exporter" version="(version=, branch=, revision=)" 114 level=info ts=2021-05-13T15:30:22.251Z caller=main.go:322 msg="Build context" context="(go=go1.16.3, user=, date=)" 115 level=info ts=2021-05-13T15:30:22.251Z caller=main.go:361 msg="Accepting StatsD Traffic" udp=localhost:8125 tcp=:9125 unixgram= 116 level=info ts=2021-05-13T15:30:22.251Z caller=main.go:362 msg="Accepting Prometheus Requests" addr=:9102 117 level=debug ts=2021-05-13T15:30:27.811Z caller=listener.go:73 msg="Incoming line" proto=udp line=aistarget.pakftUgh.kalive.latency:1|ms 118 level=debug ts=2021-05-13T15:30:29.891Z caller=listener.go:73 msg="Incoming line" proto=udp line=aisproxy.qYyhpllR.pst.count:77|c 119 level=debug ts=2021-05-13T15:30:37.811Z caller=listener.go:73 msg="Incoming line" proto=udp line=aistarget.pakftUgh.kalive.latency:1|ms 120 level=debug ts=2021-05-13T15:30:39.892Z caller=listener.go:73 msg="Incoming line" proto=udp line=aisproxy.qYyhpllR.pst.count:78|c 121 level=debug ts=2021-05-13T15:30:47.811Z caller=listener.go:73 msg="Incoming line" proto=udp line=aistarget.pakftUgh.kalive.latency:1|ms 122 level=debug ts=2021-05-13T15:30:49.892Z caller=listener.go:73 msg="Incoming line" proto=udp line=aisproxy.qYyhpllR.pst.count:79|c 123 ... 124 ``` 125 126 Finally, point any available Prometheus instance to poll the listening port - **9102** by default - of the exporter. 127 128 Note that the two listening ports mentioned - StatsD port **8125** and Prometheus port **9102** - are both configurable via the exporter's command line. To see all supported options, run: 129 130 ```console 131 $ statsd_exporter --help 132 ```