qaoru / s3_exporter

Exports Prometheus metrics about S3 buckets and objects

AWS S3 Exporter

This exporter provides metrics for AWS S3 bucket objects by querying the API with a given bucket and prefix and constructing metrics based on the returned objects.

I find it useful for ensuring that backup jobs and batch uploads are functioning by comparing the growth in size/number of objects over time, or comparing the last modified date to an expected value.

Building

make

Running

./s3_exporter <flags>

You can query a bucket and prefix combination by supplying them as parameters to /probe:

curl localhost:9340/probe?bucket=some-bucket&prefix=some-folder/some-file.txt

AWS Credentials

The exporter creates an AWS session without any configuration. You must specify credentials yourself as documented here.

Remember, if you want to load credentials from ~/.aws/config then you need to to set:

export AWS_SDK_LOAD_CONFIG=true

Docker

docker pull ribbybibby/s3-exporter

You will need to supply AWS credentials to the container, as mentioned in the previous section, either by setting the appropriate environment variables with -e, or by mounting your ~/.aws/ directory with -v.

# Environment variables
docker run -p 9340:9340 -e AWS_ACCESS_KEY_ID=<value> -e AWS_SECRET_ACCESS_KEY=<value> -e AWS_REGION=<value> s3-exporter:latest <flags>
# Mounted volume
docker run -p 9340:9340 -e AWS_SDK_LOAD_CONFIG=true -e HOME=/ -v $HOME/.aws:/.aws s3-exporter:latest <flags>

Flags

  -h, --help                     Show context-sensitive help (also try --help-long and --help-man).
      --web.listen-address=":9340"
                                 Address to listen on for web interface and telemetry.
      --web.metrics-path="/metrics"
                                 Path under which to expose metrics
      --web.probe-path="/probe"  Path under which to expose the probe endpoint
      --web.discovery-path="/discovery"
                                 Path under which to expose service discovery
      --s3.endpoint-url=""       Custom endpoint URL
      --s3.disable-ssl           Custom disable SSL
      --s3.force-path-style      Custom force path style
      --log.level="info"         Only log messages with the given severity or above. Valid levels: [debug, info, warn, error, fatal]
      --log.format="logger:stderr"
                                 Set the log target and format. Example: "logger:syslog?appname=bob&local=7" or "logger:stdout?json=true"
      --version                  Show application version.

Flags can also be set as environment variables, prefixed by S3_EXPORTER_. For example: S3_EXPORTER_S3_ENDPOINT_URL=http://s3.example.local.

Metrics

Metric	Meaning	Labels
s3_biggest_object_size_bytes	The size of the largest object.	bucket, prefix
s3_common_prefixes	A count of all the keys between the prefix and the next occurrence of the string specified by the delimiter	bucket, prefix, delimiter
s3_last_modified_object_date	The modification date of the most recently modified object.	bucket, prefix
s3_last_modified_object_size_bytes	The size of the object that was modified most recently.	bucket, prefix
s3_list_duration_seconds	The duration of the ListObjects operation	bucket, prefix, delimiter
s3_list_success	Did the ListObjects operation complete successfully?	bucket, prefix, delimiter
s3_objects_size_sum_bytes	The sum of the size of all the objects.	bucket, prefix
s3_objects	The total number of objects.	bucket, prefix

Common prefixes

Rather than generating metrics for the objects with a particular prefix, you can set the delimiter parameter to produce a count of all the keys between the prefix and the next occurrence of the given delimiter.

For instance:

$ curl 'localhost:9340/probe?bucket=registry-bucket&prefix=docker/registry/v2/blobs/sha256/&delimiter=/'
# HELP s3_common_prefixes A count of all the keys between the prefix and the next occurrence of the string specified by the delimiter
# TYPE s3_common_prefixes gauge
s3_common_prefixes{bucket="registry-bucket",delimiter="/",prefix="docker/registry/v2/blobs/sha256/"} 133
# HELP s3_list_duration_seconds The total duration of the list operation
# TYPE s3_list_duration_seconds gauge
s3_list_duration_seconds{bucket="registry-bucket",delimiter="/",prefix="docker/registry/v2/blobs/sha256/"} 0.921488535
# HELP s3_list_success If the ListObjects operation was a success
# TYPE s3_list_success gauge
s3_list_success{bucket="registry-bucket",delimiter="/",prefix="docker/registry/v2/blobs/sha256/"} 1

See this page for more information.

Prometheus

Configuration

You can pass the params to a single instance of the exporter using relabelling, like so:

scrape_configs:
  - job_name: "s3"
    metrics_path: /probe
    static_configs:
      - targets:
          - bucket=stuff;prefix=thing.txt;
          - bucket=other-stuff;prefix=another-thing.gif;
    relabel_configs:
      - source_labels: [__address__]
        regex: "^bucket=(.*);prefix=(.*);$"
        replacement: "${1}"
        target_label: "__param_bucket"
      - source_labels: [__address__]
        regex: "^bucket=(.*);prefix=(.*);$"
        replacement: "${2}"
        target_label: "__param_prefix"
      - target_label: __address__
        replacement: 127.0.0.1:9340 # S3 exporter.

Service Discovery

Rather than defining a static list of buckets you can use the /discovery endpoint in conjunction with HTTP service discovery to discover all the buckets the exporter has access to.

This should be all the config required to successfully scrape every bucket:

scrape_configs:
  - job_name: "s3"
    metrics_path: /probe
    http_sd_configs:
      - url: http://127.0.0.1:9340/discovery

Use relabel_configs to select the buckets you want to scrape:

scrape_configs:
  - job_name: "s3"
    metrics_path: /probe
    http_sd_configs:
      - url: http://127.0.0.1:9340/discovery
    relabel_configs:
      # Keep buckets that start with example-
      - source_labels: [__param_bucket]
        action: keep
        regex: ^example-.*

The prefix can be set too, but be mindful that this will apply to all buckets:

scrape_configs:
  - job_name: "s3"
    metrics_path: /probe
    http_sd_configs:
      - url: http://127.0.0.1:9340/discovery
    params:
      prefix: ["thing.txt"]

Example Queries

Return series where the last modified object date is more than 24 hours ago:

(time() - s3_last_modified_object_date) / 3600 > 24