Skip to content

Latest commit

 

History

History
829 lines (662 loc) · 24.9 KB

configuration.md

File metadata and controls

829 lines (662 loc) · 24.9 KB
Raw

Configuration

Basic configuration

The only required option to run the collector is the --config option that points to the configuration file.

otelcol-sumo --config config.yaml

For all the available command line options, see Command-line configuration options.

The file config.yaml is a regular OpenTelemetry Collector configuration file that contains a pipeline with some receivers, processors and exporters. If you are new to OpenTelemetry Collector, you can familiarize yourself with the terms reading the upstream documentation.

The primary components that make it easy to send data to Sumo Logic are the Sumo Logic Exporter and the Sumo Logic Extension.

To create the installation_token please follow this documenation.

Here's a starting point for the configuration file that you will want to use:

exporters:
  sumologic:

extensions:
  sumologic:
    installation_token: ${SUMOLOGIC_INSTALLATION_TOKEN}

receivers:
  ... # fill in receiver configurations here

service:
  extensions: [sumologic]
  pipelines:
    logs:
      receivers: [...] # fill in logs receiver names here
      exporters: [sumologic]
    metrics:
      receivers: [...] # fill in metrics receiver names here
      exporters: [sumologic]
    traces:
      receivers: [...] # fill in trace receiver names here
      exporters: [sumologic]

The Sumo Logic exporter automatically detects the Sumo Logic extension if it's added in the service.extensions property and uses it as the authentication provider to connect and send data to the Sumo Logic backend.

You add the receivers for the data you want to be collected and put them together in one pipeline. You can of course also add other components according to your needs - extensions, processors, other exporters etc.

Let's look at some examples for configuring logs, metrics and traces to be sent to Sumo, and after that let's put that all together.

IMPORTANT NOTE: It is recommended to limit access to the configuration file as it contains sensitive information. You can change access permissions to the configuration file using:

chmod 640 config.yaml

Basic configuration for logs

To send logs from local files, use the Filelog Receiver.

Example configuration:

exporters:
  sumologic:
    sending_queue:
      enabled: true
      storage: file_storage

extensions:
  file_storage:
    directory: .
  sumologic:
    installation_token: ${SUMOLOGIC_INSTALLATION_TOKEN}

receivers:
  filelog:
    include:
    - /var/log/myservice/*.log
    - /other/path/**/*.txt
    include_file_name: false
    include_file_path_resolved: true
    operators:
    - type: move
      from: attributes["log.file.path_resolved"]
      to: resource["log.file.path_resolved"]
    start_at: beginning

service:
  extensions: [file_storage, sumologic]
  pipelines:
    logs:
      receivers: [filelog]
      exporters: [sumologic]

Adding the File Storage extension allows the Filelog receiver to persist the position in the files it reads between restarts.

See section below on Collecting logs from files for details on configuring the Filelog receiver.

Basic configuration for metrics

Sumo Logic Distribution for OpenTelemetry Collector uses the Host Metrics Receiver to ingest metrics.

Here's a minimal config.yaml file that sends the host's memory metrics to Sumo Logic:

exporters:
  sumologic:

extensions:
  sumologic:
    installation_token: ${SUMOLOGIC_INSTALLATION_TOKEN}

receivers:
  hostmetrics:
    collection_interval: 30s
    scrapers:
      cpu:
      memory:

service:
  extensions: [sumologic]
  pipelines:
    metrics:
      receivers: [hostmetrics]
      exporters: [sumologic]

Basic configuration for traces

Use the OTLP Receiver to ingest traces.

Example configuration:

exporters:
  sumologic:

extensions:
  sumologic:
    installation_token: ${SUMOLOGIC_INSTALLATION_TOKEN}

receivers:
  otlp:
    protocols:
      grpc:

service:
  extensions: [sumologic]
  pipelines:
    traces:
      receivers: [otlp]
      exporters: [sumologic]

Putting it all together

Here's an example configuration file that collects all the signals - logs, metrics and traces.

exporters:
  sumologic:
    sending_queue:
      enabled: true
      storage: file_storage

extensions:
  file_storage:
    directory: .
  sumologic:
    installation_token: ${SUMOLOGIC_INSTALLATION_TOKEN}

receivers:
  filelog:
    include:
    - /var/log/myservice/*.log
    - /other/path/**/*.txt
    include_file_name: false
    include_file_path_resolved: true
    operators:
    - type: move
      from: attributes["log.file.path_resolved"]
      to: resource["log.file.path_resolved"]
    start_at: beginning
  hostmetrics:
    collection_interval: 30s
    scrapers:
      cpu:
      memory:
  otlp:
    protocols:
      grpc:

service:
  extensions: [file_storage, sumologic]
  pipelines:
    logs:
      receivers: [filelog]
      exporters: [sumologic]
    metrics:
      receivers: [hostmetrics]
      exporters: [sumologic]
    traces:
      receivers: [otlp]
      exporters: [sumologic]

Authentication

To send data to Sumo Logic you need to configure the sumologicextension with credentials and define it (the extension) in the same service as the sumologicexporter is defined so that it's used as an auth extension.

The following configuration is a basic example to collect CPU load metrics using the Host Metrics Receiver and send them to Sumo Logic:

extensions:
  sumologic:
    installation_token: ${SUMOLOGIC_INSTALLATION_TOKEN}
    collector_name: ${MY_COLLECTOR_NAME}

receivers:
  hostmetrics:
    collection_interval: 30s
    scrapers:
      load:

exporters:
  sumologic:

service:
  extensions: [sumologic]
  pipelines:
    metrics:
      receivers: [hostmetrics]
      exporters: [sumologic]

For a list of all the configuration options for sumologicextension refer to this documentation.

Running the collector as systemd service

The credentials are stored on local filesystem (by default in $HOME/.sumologic-otel-collector) to be reused when collector gets restarted. However, systemd services are often run as users without a home directory, so keep in mind that to store credentials either the user needs a home directory or the store location should be explicitely changed.

More information about this feature can be found in the extension's documentation.

Using multiple Sumo Logic extensions

If you want to register multiple collectors and/or send data to mutiple Sumo Logic accounts, mutiple sumologicextensions can be defined within the pipeline and used in exporter definitions.

In this case, you need to specify a custom authenticator name that points to the correct extension ID.

Example:

extensions:
  sumologic/custom_auth1:
    installation_token: ${SUMOLOGIC_INSTALLATION_TOKEN_1}
    collector_name: <COLLECTOR_NAME_1>

  sumologic/custom_auth2:
    installation_token: ${SUMOLOGIC_INSTALLATION_TOKEN_2}
    collector_name: <COLLECTOR_NAME_2>

receivers:
  hostmetrics:
    collection_interval: 30s
    scrapers:
      load:
  filelog:
    include: [ "**.log" ]

exporters:
  sumologic/custom1:
    auth:
      authenticator: sumologic/custom_auth1
  sumologic/custom2:
    auth:
      authenticator: sumologic/custom_auth2

service:
  extensions: [sumologic/custom_auth1, sumologic/custom_auth2]
  pipelines:
    metrics/1:
      receivers: [hostmetrics]
      exporters: [sumologic/custom1]
    logs/1:
      receivers: [filelog]
      exporters: [sumologic/custom2]

Persistence

When using the Sumo Logic exporter, it is recommended to store its state in a persistent storage to prevent loss of data buffered in the exporter between restarts.

To do that, add the File Storage extension to the configuration and configure the exporter to use persistent queue with the sending_queue.enabled: true and sending_queue.storage: <storage_name> flags.

Here's an example configuration:

exporters:
  sumologic:
    sending_queue:
      enabled: true
      storage: file_storage

extensions:
  file_storage:
    directory: .
  sumologic:
    installation_token: ${SUMOLOGIC_INSTALLATION_TOKEN}

receivers:
  hostmetrics:
    collection_interval: 3s
    scrapers:
      load:

service:
  extensions:
  - file_storage
  - sumologic
  pipelines:
    metrics:
      exporters:
      - sumologic
      receivers:
      - hostmetrics

Collecting logs from files

The Filelog Receiver tails and parses logs from files using the opentelemetry-log-collection library.

The following is a basic configuration for the Filelog Receiver:

receivers:
  filelog:
    include:
    - /var/log/myservice/*.log
    - /other/path/**/*.txt
    include_file_name: false
    include_file_path_resolved: true
    operators:
    - type: move
      from: attributes["log.file.path_resolved"]
      to: resource["log.file.path_resolved"]

The include_file_name: false prevents the receiver from adding log.file.name attribute to the logs. Instead, we are using include_file_path_resolved: true, which adds a log.file.path_resolved attribute to the logs that contains the whole path of the file, as opposed to just the name of the file. What's more, the log.file.path_resolved attribute is automatically recognized by the sumologicexporter and translated to _sourceName attribute in Sumo Logic. We just need the move operator to move the attribute from record level to resource level.

Keeping track of position in files

By default, the Filelog receiver watches files starting at their end (start_at: end is the default), so nothing will be read after the otelcol process starts until new data is added to the files. To change this, add start_at: beginning to the receiver's configuration. To prevent the receiver from reading the same data over and over again on each otelcol restart, also add the File Storage extension that will allow Filelog receiver to persist the current position in watched files between otelcol restarts. Here's an example of such configuration:

extensions:
  file_storage:
    directory: .

receivers:
  filelog:
    include:
    - /var/log/myservice/*.log
    - /other/path/**/*.txt
    include_file_name: false
    include_file_path_resolved: true
    operators:
    - type: move
      from: attributes["log.file.path_resolved"]
      to: resource["log.file.path_resolved"]
    start_at: beginning

For more details, see the Filelog Receiver documentation.

Parsing JSON logs

Filelog Receiver with json_parser operator can be used for parsing JSON logs. The json_parser operator parses the string-type field selected by parse_from as JSON (by default parse_from is set to $body which indicates the whole log record).

For example when logs has following form in the file:

{"message": "{\"key\": \"val\"}"}

then configuration to extract JSON which is represented as string ({\"key\": \"val\"}) has following form:

receivers:
  filelog:
    include:
    - /log/path/**/*.log
    operators:
      - type: json_parser   # this parses log line as JSON
      - type: json_parser   # this parses string under 'message' key as JSON
        parse_from: message

and the parsed log entry can be observed in logging exporter's output as:

2022-02-24T10:23:37.809Z        INFO    loggingexporter/logging_exporter.go:69  LogsExporter    {"#logs": 1}
2022-02-24T10:23:37.809Z        DEBUG   loggingexporter/logging_exporter.go:79  ResourceLog #0
Resource SchemaURL:
InstrumentationLibraryLogs #0
InstrumentationLibraryMetrics SchemaURL:
InstrumentationLibrary
LogRecord #0
Timestamp: 2022-02-24 10:23:37.714661255 +0000 UTC
Severity:
ShortName:
Body: {
     -> key: STRING(val)
}
Attributes:
     -> log.file.name: STRING(example.log)
Trace ID:
Span ID:
Flags: 0

Example configuration with example log can be found in /examples/logs_json/ directory.

Setting source category

For many Sumo Logic customers the source category is a crucial piece of metadata that allows to differentiate data coming from different sources. This section describes how to configure the Sumo Logic Distribution of OpenTelemetry to decorate data with this attribute.

To decorate all the data from the collector with the same source category, set the source_category property on the source processor like this:

processors:
  source:
    source_category: my/source/category

You can also use the source templates to use other attributes' values to compose source category, like this:

processors:
  source:
    source_category: "%{k8s.namespace.name}/%{k8s.pod.pod_name}"

However, please bear in mind that the source processor is going to be removed in the future.

If you want data from different sources to have different source categories, you'll need to set a resource attribute named _sourceCategory earlier in the pipeline. See below for examples on how to do this in various scenarios.

Setting source category on logs from files

The Filelog receiver allows to set attributes on received data. Here's an example on how to attach different source categories to logs from different files with the Filelog receiver:

exporters:
  sumologic:

receivers:

  filelog/apache:
    include:
    - /logs/from/apache.txt
    resource:
      _sourceCategory: apache-logs

  filelog/nginx:
    include:
    - /logs/from/nginx.txt
    resource:
      _sourceCategory: nginx-logs

service:
  pipelines:
    logs:
      exporters:
      - sumologic
      receivers:
      - filelog/apache
      - filelog/nginx

Setting source category on Prometheus metrics

The Prometheus receiver uses the same config as the full-blown Prometheus, so you can use its relabel config to create the _sourceCategory label.

exporters:
  sumologic:

receivers:
  prometheus:
    config:
      scrape_configs:

      - job_name: mysql-metrics
        relabel_configs:
        - replacement: db-metrics
          target_label: _sourceCategory
        static_configs:
        - targets:
          - "0.0.0.0:9104"

      - job_name: otelcol-metrics
        relabel_configs:
        - source_labels:
          - job
          target_label: _sourceCategory
        static_configs:
        - targets:
          - "0.0.0.0:8888"

service:
  pipelines:
    metrics:
      exporters:
      - sumologic
      receivers:
      - prometheus

The first example creates a _sourceCategory label with a hardcoded value of db-metrics.

The second example creates a _sourceCategory label by copying to it the value of Prometheus' job label, which contains the name of the job - in this case, otelcol-metrics.

Setting source category with the Resource processor

If the receiver does not allow to set attributes on the received data, use the Resource processor to add the _sourceCategory attribute later in the pipeline. Here's an example on how to get statsd metrics from two different apps to have different source categories.

exporters:
  sumologic:

processors:

  resource/one-source-category:
    attributes:
    - action: insert
      key: _sourceCategory
      value: one-app

  resource/another-source-category:
    attributes:
    - action: insert
      key: _sourceCategory
      value: another-app

receivers:

  statsd/one-app:
    endpoint: "localhost:8125"

  statsd/another-app:
    endpoint: "localhost:8126"

service:
  pipelines:
    metrics/one-app:
      exporters:
      - sumologic
      processors:
      - resource/one-app
      receivers:
      - statsd/one-app

    metrics/another-app:
      exporters:
      - sumologic
      processors:
      - resource/another-app
      receivers:
      - statsd/another-app

Setting source host

You can use the source processor's source_host property to set the Sumo Logic source host attribute to a static value like this:

processors:
  source:
    source_host: my-host-name

But this is most likely not what you want. You'd rather have the collector retrieve the name of the host from the operating system, instead of needing to manually hardcode it in the config.

This is what the Resource Detection processor does. Use its built-in system detector to set the OpenTelemetry standard host.name resource attribute to the name of the host that the collector is running on. After that is set, you need to add the _sourceHost attribute with the value from the host.name attribute.

exporters:
  sumologic:

processors:
  resource/add-source-host:
    attributes:
    - action: insert
      key: _sourceHost
      from_attribute: host.name
    # Optionally, delete the original attributes created by the Resource Detection processor.
    - action: delete
      key: host.name
    - action: delete
      key: os.type

  resourcedetection/detect-host-name:
    detectors:
    - system
    system:
      hostname_sources:
      - os

receivers:
  hostmetrics:
    scrapers:
      memory:

service:
  pipelines:
    metrics:
      exporters:
      - sumologic
      processors:
      - resourcedetection/detect-host-name
      - resource/add-source-host
      receivers:
      - hostmetrics

Make sure to put the Resource Detection processor before the Resource processor in the pipeline so that the host.name attribute is already set in the resource processor.

Only the first Resource processor's action is required to correctly set the _sourceHost attribute. The other two actions perform an optional metadata cleanup - they delete the unneeded attributes.

Command-line configuration options

Usage:
  otelcol-sumo [flags]

Flags:
      --config -config=file:/path/to/first --config=file:path/to/second   Locations to the config file(s), note that only a single pattern can be set per flag entry e.g. -config=file:/path/to/single_file --config="glob:path/with/glob/*". (default [])
      --feature-gates Flag                                                Comma-delimited list of feature gate identifiers. Prefix with '-' to disable the feature.  '+' or no prefix will enable the feature.
  -h, --help                                                              help for otelcol-sumo
      --set stringArray                                                   Set arbitrary component config property. The component has to be defined in the config file and the flag has a higher precedence. Array config properties are overridden and maps are joined, note that only a single (first) array property can be set e.g. -set=processors.attributes.actions.key=some_key. Example --set=processors.batch.timeout=2s (default [])
  -v, --version                                                           version for otelcol-sumo

Proxy Support

Exporters leverage the HTTP communication and respect the following proxy environment variables:

  • HTTP_PROXY
  • HTTPS_PROXY
  • NO_PROXY

You may either export proxy environment variables locally e.g.

export FTP_PROXY=<PROXY-ADDRESS>:<PROXY-PORT>
export HTTP_PROXY=<PROXY-ADDRESS>:<PROXY-PORT>
export HTTPS_PROXY=<PROXY-ADDRESS>:<PROXY-PORT>

or make them available globally for all users, e.g.

tee -a /etc/profile << END
export FTP_PROXY=<PROXY-ADDRESS>:<PROXY-PORT>
export HTTP_PROXY=<PROXY-ADDRESS>:<PROXY-PORT>
export HTTPS_PROXY=<PROXY-ADDRESS>:<PROXY-PORT>
END

Keeping Prometheus format using OTLP exporter

In order to keep the Prometheus compatible metric names using OTLP exporting format, the Metrics Transform Processor can be used.

Please see the following example of replacing last period char with underscore:

processors:
  metricstransform:
    transforms:
      ## Replace last period char in metric name with underscore
      - include: ^(.*)\.([^\.]*)$$
        match_type: regexp
        action: update
        new_name: $${1}_$${2}
# ...
service:
  pipelines:
    metrics:
      receivers:
        - telegraf
      processors:
        - metricstransform
      exporters:
        - sumologic
# ...