vtgen

vtgen

vtgen is a trace data generator. It reads the OTLP request body in testdata/testdata.bin, modifies the trace_id, start_time and end_time, and sends them to OTLP trace endpoint (/v1/traces).

Usage

vtgen can be used for:

1. Generating identical data for different targets, which is useful for query performance benchmarking of different storage backends against the same dataset.
2. Generating different data at the same rate (on a best-effort basis) for different targets, which is useful for data ingestion performance benchmarking of different storage backends.

You can build vtgen via the following commands:

# build via go build
make vtgen
# build via Docker
make vtgen-prod

These commands will generate vtgen(-prod) binary in bin folder.

vtgen MUST be run from the root path of VictoriaTraces repository, to load the test data correctly from app/vtgen/testdata.

To send identical data to different targets:

./bin/vtgen -addrs=http://example-url1:port/v1/traces,http://example-url2:port/insert/opentelemetry/v1/traces

The performance of different targets will affect each other, as vtgen generates data and makes HTTP requests to them one by one.

To send (potentially) different data to different addresses at the same rate, simply run multiple vtgen with different HTTP listening port:

./bin/vtgen -addrs=http://example-url1:port/v1/traces -httpListenAddr=0.0.0.0:8080
./bin/vtgen -addrs=http://example-url2:port/insert/opentelemetry/v1/traces -httpListenAddr=0.0.0.0:8081

Metrics

vtgen exposes some metrics to help tracking the ingestion performance. Here's the example:

vt_gen_http_request_duration_seconds_bucket{path="http://example-url1:port/v1/traces",vmrange="7.743e-05...8.799e-05"} 8
vt_gen_http_request_duration_seconds_bucket{path="http://example-url1:port/v1/traces",vmrange="3.162e-03...3.594e-03"} 1
vt_gen_http_request_duration_seconds_sum{path="http://example-url1:port/v1/traces"} 0.17367758299999986
vt_gen_http_request_duration_seconds_count{path="http://example-url1:port/v1/traces"} 667
vt_gen_http_request_error_count{path="http://example-url1:port/v1/traces"} 667

• vt_gen_http_request_duration_seconds_bucket/_sum/_count is a histogram for each address.
• vt_gen_http_request_error_count is a counter for each address.

Command-line flags

  -addrs string
        otlp trace export endpoints, split by ",".
  -authorizations string
        authorization header for each -addrs, split by ",".
  -httpListenAddr string
        http listen address for pprof and metrics. (default "0.0.0.0:8080")
  -logEvery10k int
        how many trace id should be logged for every 10000 traces by each worker. (default 2)
  -rate int
        spans per second. (default 10000)
  -worker int
        number of workers. (default 4)