This is the new MongoDB exporter implementation that handles ALL metrics exposed by MongoDB monitoring commands. This new implementation loops over all the fields exposed in diagnostic commands and tries to get data from them.
Currently, these metric sources are implemented:
- $collStats
- $indexStats
- getDiagnosticData
- replSetGetStatus
- serverStatus
The exporter works with Percona Server for MongoDB and MongoDB Community or Enterprise Edition versions 4.4 and newer. Older versions might also work but are not tested anymore.
The old 0.1x.y version (ex master
branch) has been moved to the release-0.1x
branch.
A flag, --compatible-mode
, which exposes metrics with 0.1x compatible metric
names has been implemented which simplifies migration from the old version to
the current version.
The build process uses the dockerized version of goreleaser so you don't need to install Go.
Just run make release
and the new binaries will be generated under the build directory.
├── build
│ ├── config.yaml
│ ├── mongodb_exporter_7c73946_checksums.txt
│ ├── mongodb_exporter-7c73946.darwin-amd64.tar.gz
│ ├── mongodb_exporter-7c73946.linux-amd64.tar.gz
│ ├── mongodb_exporter_darwin_amd64
│ │ └── mongodb_exporter <--- MacOS binary
│ └── mongodb_exporter_linux_amd64
│ └── mongodb_exporter <--- Linux binary
If you built the exporter using the method mentioned in the previous section, the generated binaries are in mongodb_exporter_linux_amd64/mongodb_exporter
or mongodb_exporter_darwin_amd64/mongodb_exporter
A docker image is available on the official percona repository.
# with podman
podman run -d -p 9216:9216 percona/mongodb_exporter:0.40 --mongodb.uri=mongodb://127.0.0.1:17001
# with docker
docker run -d -p 9216:9216 percona/mongodb_exporter:0.40 --mongodb.uri=mongodb://127.0.0.1:17001
Connecting user should have sufficient rights to query needed stats:
{
"role":"clusterMonitor",
"db":"admin"
},
{
"role":"read",
"db":"local"
}
More info about roles in MongoDB documentation.
mongodb_exporter_linux_amd64/mongodb_exporter --mongodb.uri=mongodb://127.0.0.1:17001
You can supply the mongodb user/password direct in the --mongodb.uri=
like --mongodb.uri=mongodb://user:pass@127.0.0.1:17001
, you can also supply the mongodb user/password with --mongodb.user=
, --mongodb.password=
but the user and password info will be leaked via ps
or top
command, for security issue, you can use MONGODB_USER
and MONGODB_PASSWORD
env variable to set user/password for given uri
MONGODB_USER=XXX MONGODB_PASSWORD=YYY mongodb_exporter_linux_amd64/mongodb_exporter --mongodb.uri=mongodb://127.0.0.1:17001 --mongodb.collstats-colls=db1.c1,db2.c2
# or
export MONGODB_USER=XXX
export MONGODB_PASSWORD=YYY
mongodb_exporter_linux_amd64/mongodb_exporter --mongodb.uri=mongodb://127.0.0.1:17001 --mongodb.collstats-colls=db1.c1,db2.c2
You can run the exporter specifying multiple URIs, devided by a comma in --mongodb.uri option or MONGODB_URI environment variable in order to monitor multiple mongodb instances with the a single mongodb_exporter instance.
--mongodb.uri=mongodb://user:pass@127.0.0.1:27017/admin,mongodb://user2:pass2@127.0.0.1:27018/admin
In this case you can use the /scrape endpoint with the target parameter to retreive the specified tartget's metrics. When querying the data you can use just mongodb://host:port in the target parameter without other parameters and, of course without host credentials
GET /scrape?target=mongodb://127.0.0.1:27018
If your URI is prefixed by mongodb:// or mongodb+srv:// schema, any host not prefixed by it after comma is being treated as part of a cluster rather then as a standalone host. Thus clusters and standalone hosts can be combined like this:
--mongodb.uri=mongodb+srv://user:pass@host1:27017,host2:27017,host3:27017/admin,mongodb://user2:pass2@host4:27018/admin
--mongodb.collstats-colls
receives a list of databases and collections to monitor using collstats.
Usage example: --mongodb.collstats-colls=database1.collection1,database2.collection2
mongodb_exporter_linux_amd64/mongodb_exporter --mongodb.uri=mongodb://127.0.0.1:17001 --mongodb.collstats-colls=db1.c1,db2.c2
When compatibility mode is enabled by the --compatible-mode
, the exporter will expose all new metrics with the new naming and labeling schema and at the same time will expose metrics in the version 1 compatible way.
For example, if compatibility mode is enabled, the metric mongodb_ss_wt_log_log_bytes_written
(new format)
# HELP mongodb_ss_wt_log_log_bytes_written serverStatus.wiredTiger.log.
# TYPE mongodb_ss_wt_log_log_bytes_written untyped
mongodb_ss_wt_log_log_bytes_written 2.6208e+06
will be also exposed as mongodb_mongod_wiredtiger_log_bytes_total
with the unwritten
label.
HELP mongodb_mongod_wiredtiger_log_bytes_total mongodb_mongod_wiredtiger_log_bytes_total
# TYPE mongodb_mongod_wiredtiger_log_bytes_total untyped
mongodb_mongod_wiredtiger_log_bytes_total{type="unwritten"} 2.6208e+06
--collector.profile
To collect metrics, you need to enable the profiler in MongoDB:
Usage example: db.setProfilingLevel(2)
Level | Description |
---|---|
0 | The profiler is off and does not collect any data. This is the default profiler level. |
1 | The profiler collects data for operations that take longer than the value of slowms or that match a filter.When a filter is set:
|
2 | The profiler collects data for all operations. |
When shard metrics collection is enabled by --collector.shards
, the exporter will expose metrics related to sharded Mongo.
Example, if shards collector is enabled:
# HELP mongodb_shards_collection_chunks_count sharded collection chunks.
# TYPE mongodb_shards_collection_chunks_count counter
mongodb_shards_collection_chunks_count{collection="system.sessions",database="config",shard="rs1"} 250
mongodb_shards_collection_chunks_count{collection="system.sessions",database="config",shard="rs2"} 250
You can see shard name, it's collection, database and count.
The exporter sets some topology labels in all metrics. The labels are:
- cl_role: Cluster role according to this table:
Server type | Label |
---|---|
mongos | mongos |
regular instance (primary or secondary) | shardsvr |
arbiter | shardsvr |
standalone | (empty string) |
- cl_id: Cluster ID
- rs_nm: Replicaset name
- rs_state: Replicaset state is an integer from
getDiagnosticData()
->replSetGetStatus.myState
. Check the official documentation for details on replicaset status values.
See the Reference Guide for details on using the exporter.
Refer to the Contribution Guide.