This is a builtin exporter from OpenSearch to Prometheus based on the code of Vincent Van Hollebeke and others at https://github.com/vvanholl/elasticsearch-prometheus-exporter . It collects all relevant metrics and makes them available to Prometheus via the OpenSearch REST API.
Currently, the available metrics are:
- Cluster status
- Nodes status:
- JVM
- Indices (global)
- Transport
- HTTP
- Scripts
- Process
- Operating System
- File System
- Circuit Breaker
- Indices status
- Cluster settings
| OpenSearch | Plugin | Release date |
|---|---|---|
| 1.0.0 | 1.0.0 | Jun 24, 2021 |
| 1.1.0 | 1.1.0 | Oct 08, 2021 |
| 1.2.0 | 1.2.0 | Nov 27, 2021 |
| 1.2.2 | 1.2.2 | Dec 19, 2021 |
| 1.2.3 | 1.2.3 | Dec 24, 2021 |
./bin/opensearch-plugin install -b https://github.com/aparo/opensearch-prometheus-exporter/releases/download/1.2.3/prometheus-exporter-1.2.3.zip
Do not forget to restart the node after the installation!
Note that the plugin needs the following special permissions:
- java.lang.RuntimePermission accessClassInPackage.sun.misc
- java.lang.RuntimePermission accessDeclaredMembers
- java.lang.reflect.ReflectPermission suppressAccessChecks
If you have a lot of indices and think this data is irrelevant, you can disable in the main configuration file:
prometheus.indices: false
To disable exporting cluster settings use:
prometheus.cluster.settings: false
These settings can be also updated dynamically.
./bin/opensearch-plugin remove prometheus-exporter
Do not forget to restart the node after installation!
Metrics are directly available at:
http://<your-opensearch-host>:9200/_prometheus/metrics
As a sample result, you get:
# HELP es_process_mem_total_virtual_bytes Memory used by ES process
# TYPE es_process_mem_total_virtual_bytes gauge
es_process_mem_total_virtual_bytes{cluster="develop",node="develop01",} 3.626733568E9
# HELP es_indices_indexing_is_throttled_bool Is indexing throttling ?
# TYPE es_indices_indexing_is_throttled_bool gauge
es_indices_indexing_is_throttled_bool{cluster="develop",node="develop01",} 0.0
# HELP es_jvm_gc_collection_time_seconds Time spent for GC collections
# TYPE es_jvm_gc_collection_time_seconds counter
es_jvm_gc_collection_time_seconds{cluster="develop",node="develop01",gc="old",} 0.0
es_jvm_gc_collection_time_seconds{cluster="develop",node="develop01",gc="young",} 0.0
# HELP es_indices_requestcache_memory_size_bytes Memory used for request cache
# TYPE es_indices_requestcache_memory_size_bytes gauge
es_indices_requestcache_memory_size_bytes{cluster="develop",node="develop01",} 0.0
# HELP es_indices_search_open_contexts_number Number of search open contexts
# TYPE es_indices_search_open_contexts_number gauge
es_indices_search_open_contexts_number{cluster="develop",node="develop01",} 0.0
# HELP es_jvm_mem_nonheap_used_bytes Memory used apart from heap
# TYPE es_jvm_mem_nonheap_used_bytes gauge
es_jvm_mem_nonheap_used_bytes{cluster="develop",node="develop01",} 5.5302736E7
...
On your Prometheus servers, configure a new job as usual.
For example, if you have a cluster of 3 nodes:
- job_name: opensearch
scrape_interval: 10s
metrics_path: "/_prometheus/metrics"
static_configs:
- targets:
- node1:9200
- node2:9200
- node3:9200Of course, you could use the service discovery service instead of a static config.
Just keep in mind that metrics_path must be /_prometheus/metrics, otherwise Prometheus will find no metric.
If you enable security in OpenSearch, remember to set up the credentials also in Prometheus configuration.
This plugin mainly uses the Prometheus JVM Client.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.