| layout | title | nav_order | parent | redirect_from | grand_parent | |
|---|---|---|---|---|---|---|
default |
Command flags |
51 |
Command reference |
|
OpenSearch Benchmark Reference |
OpenSearch Benchmark uses command line flags to change Benchmark's behavior. Not all flags can be used with each command. To find out which flags are supported by a specific command, enter opensearch-benchmark <command> --h.
All command flags are added to a command using the following syntax:
opensearch-benchmark <command> --<command-flag>Flags that accept comma-separated values, such --telemetry, can also accept a JSON array. This can be defined by passing a file path ending in .json or inline as a JSON string.
- Comma-seperated values:
opensearch-benchmark ... --test-procedure="ingest-only,search-aggregations" - JSON file:
opensearch-benchmark ... --workload-params="params.json" - JSON inline string:
opensearch-benchmark ... --telemetry='["node-stats", "recovery-stats"]'
This can be either a directory that contains a workload.json file or a .json file with an arbitrary name that contains a workload specification. --workload-path and --workload-repository as well as --workload are mutually exclusive.
This defines the repository from which OpenSearch Benchmark loads workloads. --workload-path and --workload-repository as well as --workload are mutually exclusive.
Defines a specific revision from the workload source tree that OpenSearch Benchmark should use.
Defines the workload to use based on the workload's name. You can find a list of preloaded workloads using opensearch-benchmark list workloads. --workload-path and --workload-repository as well as --workload are mutually exclusive.
Defines which variables to inject into the workload. Variables injected must be available in the workload. To see which parameters are valid in the official workloads, select the workload from the workloads repository.
Defines the test procedures to use with each workload. You can find a list of test procedures that the workload supports by specifying the workload in the info command, for example, opensearch-benchmark info --workload=<workload_name>. To look up information on a specific test procedure, use the command opensearch-benchmark info --workload=<workload_name> --test-procedure=<test-procedure>.
Defines a unique ID for the test run.
Defines a comma-separated list of test procedure tasks to run. By default, all tasks listed in a test procedure array are run.
Tests are executed in the order they are defined in test-procedure---not in the order they are defined in the command.
All task filters are case sensitive.
Defines a comma-separated list of test procedure tasks not to run.
The baseline TestExecution ID used to compare the contender TestExecution.
The TestExecution ID for the contender being compared to the baseline.
Defines the output format for the command line results, either markdown or csv. Default is markdown.
Defines the column number alignment for when the compare command outputs results. Default is right.
When provided a file path, writes the compare results to the file indicated in the path.
Determines whether or not to include the comparison in the results file.
Defines the repository from which OpenSearch Benchmark loads provision-configs and provision-config-instances.
Defines the specific Git revision in the provision-config that OpenSearch Benchmark should use.
Defines the path to the --provision-config-instance and any OpenSearch plugin configurations to use.
Downloads the specified OpenSearch distribution based on version number. For a list of released OpenSearch versions, see Version history.
Defines the repository from which the OpenSearch distribution should be downloaded. Default is release.
Defines the --provision-config-instance to use. You can view possible configuration instances by using the command opensearch-benchmark list provision-config-instances.
A comma-separated list of key-value pairs injected verbatim as variables for the provision-config-instance.
Defines a comma-separated list of host-port pairs that should be targeted if using the pipeline benchmark-only. Default is localhost:9200.
The target operating system (OS) for which the OpenSearch artifact should be downloaded. Default is the current OS.
The name of the CPU architecture for which an artifact should be downloaded.
Defines the current source code revision to use for running a benchmark test. Default is current.
This command flag can use the following options:
current: Uses the source tree's current revision based on your OpenSearch distribution.latest: Fetches the latest revision from the main branch of the source tree.
You can also use a timestamp or commit ID from the source tree. When using a timestamp, specify @ts, where "ts" is a valid ISO 8601 timestamp, for example, @2013-07-27T10:37:00Z.
Defines which OpenSearch plugins to install. By default, no plugins are installed.
Defines a comma-separated list of key-value pairs that are injected verbatim into all plugins as variables.
The major version of JDK to use.
Defines a comma-separated list of clients to use. All options are passed to the OpenSearch Python client. Default is timeout:60.
Defines a comma-separated list of hosts that coordinate loads. Default is localhost.
Enables a performance analysis of OpenSearch Benchmark's worker coordinator. Default is false.
The --pipeline option selects a pipeline to run. You can find a list of pipelines supported by OpenSearch Benchmark by running opensearch-benchmark list pipelines.
Enables the provided telemetry devices when the devices are provided using a comma-separated list. You can find a list of possible telemetry devices by using opensearch-benchmark list telemetry.
Enables setting parameters for telemetry devices. Accepts a list of comma-separated key-value pairs, each of which are delimited by a colon or a JSON file name.
Controls how OpenSearch Benchmark responds to errors. Default is continue.
You can use the following options with this command flag:
continue: Continues to run the test despite the error.abort: Aborts the test when an error occurs.
Keeps the Benchmark candidate and its index. Default is false.
When set to true, stops any OpenSearch Benchmark processes currently running and allows Benchmark to continue to run. Default is false.
Sets the path to the JSON files containing chart specifications that can be used to generate charts.
Generates the indicated chart type, either time-series or bar. Default is time-series.
The name and path used for the chart's output. Default is stdout.
Limits the number of search results for recent test runs. Default is 10.
Specifies a comma-separated list of latency percentiles to report after the workload runs. Accepts ints or floats with values between 0 and 100 inclusive. Does not accept min, median, mean, or max. Default is 50,90,99,99.9,99.99,100.
Specifies a list of throughput percentiles to report after the workload runs, in addition to min/median/mean/max which is always displayed. Like --latency-percentiles, the setting accepts ints or floats with values between 0 and 100 inclusive. Does not accept min, median, mean, or max. Default is None.
Enables randomization of values in range queries, where the values are drawn from standard value functions registered with register_standard_value_source in the workload's workload.py file.
A standard value function is a no-argument function that generates a random pair of values for a certain field, in a dict with keys "gte", "lte", and optionally "format".
If this argument is True but a search operation does not have a registered standard value function, OpenSearch Benchmark raises a SystemSetupError.
Default is False.
Sets what fraction of randomized query values can be repeated. Takes values between 0.0 and 1.0. Default is 0.3. This setting does not work when --randomization-enabled is not used.
Sets how many distinct repeatable pair values are generated for each operation when randomization is used. Default is 5000. This setting does not work when --randomization-enabled is not used.