GBench
HTTP Load Testing And Benchmarking Tool inspired by Apache Benchmark and Siege.
Requirements
You need Golang installed and ready on your system.
Installation
go install github.com/sasanrose/gbench
Usage:
Gbench has three main subcommands: exec, json and render. Exec is used, as the name suggests, to execute a benchmark based on a single url. While json subcommands runs benchmark based on a JSON configruation file and allows you to run benchmarks for different paths for a single host concurrently. This has the benefit of reproducing the same benchmark just by sharing the JSON file. Both exec and json subcommands store the result of benchmark in a JSON file (report.json in the current directy by default). And at last but not least, the render subcommand can convert the generated report to a more human readable format.
$ gbench -h
Usage:
gbench [flags]
gbench [command]
Available Commands:
exec Executes the benchmark
help Help about any command
json Executes the benchmark using json configuration
render Render the report generated by exec command
Flags:
-h, --help help for gbench
Use "gbench [command] --help" for more information about a command.
$ gbench exec -h
Usage:
gbench exec [flags]
Flags:
-c, --concurrency int Number of concurrent requests. (default 1)
--connect-timeout duration Connection timeout (0 means no timeout).
-b, --cookie string A string to be sent as raw cookie (In the format of Set-Cookie HTTP header).
-d, --data strings Sends the specified data in a request. The format should be 'key=val' or 'key1=val1&key2=val2'. This can be used multiple times.
-F, --force Force overwrite for the report file.
-H, --header strings HTTP header in format of 'key: value' or 'key: value;' or 'key;'. This can be used multiple times.
-h, --help help for exec
-o, --output string The path to store the report of benchmark. (default "./report.json")
--proxy string HTTP proxy.
-X, --request string Specify a custom HTTP method. (default "GET")
--response-timeout duration Response timeout (0 means no timeout).
-s, --status-codes ints Define what should be considered as a successful status code. (default [200,202,201])
-r, --total-requests int Number of total requests to send. (default 1)
-u, --user string Specify the user name and password to use for server authentication in the format of user:password. Currently only supports Basic Auth.
The user name and passwords are split up on the first colon, as a result it is impossible to use a colon in the user name.
$ gbench json -h
Usage:
gbench json [flags]
Flags:
-F, --force Force overwrite for the report file.
-h, --help help for json
-o, --output string The path to store the report of benchmark. (default "./report.json")
$ gbench render -h
Sample usage:
gbench render (Will use all the default values)
gbench render -i ./path/to/report.json
gbench render -i ./path/to/report.json --driver html
gbench render -i ./path/to/report.json --driver html -a 0.0.0.0 -p 7777
Usage:
gbench render [flags]
Flags:
-a, --address string Address to access the html report. (default "localhost")
-d, --driver string Driver to use for rendering the report. Accepted values are 'cli'and 'html'. (default "cli")
-h, --help help for render
-i, --input string Path to the report file. (default "./report.json")
-p, --port string Port to access the html report. (default "8080")
The following is a sample JSON config file that can be used with json subcommand. Most of the keys are based on flags of exec subcommand. The only required keys are host and paths.
{
"host": "http://localhost:8080",
"concurrency": 5,
"requests": 100,
"status-codes": [200, 201],
"user": "user:pass",
"proxy": "http://proxy:3333",
"connect-timeout": 1000000000,
"response-timeout": 5000000000,
"headers": ["X-Custome-Header: TestValue;"],
"cookie": "some-raw-cookie",
"paths": [
{
"path": "/"
},
{
"path": "/test",
"method": "post",
"user": "user:pass",
"headers": ["X-Custome-Header: TestValue;"],
"cookie": "some-raw-cookie",
"data": ["key1=val1&key2=val2", "key3=val3"]
}
]
}
Disclaimer: Gbench is still beta version. The API may change in future.
