CHttp is a tool to send HTTP requests. The tool is based on .NET8 and .NET9, for HTTP/3 uses msquic. Linux dotnet tool installations should get libmsquic package.
To install as a dotnet tool:
dotnet tool install -g LaDeak.CHttpExec --prerelease
Download latest executables from this GitHub repository's Releases page.
Download from Visual Studio Code marketplace as CHttp.
Run command as:
chttp --method GET --uri https://localhost:5001
or
chttp.exe --method GET --uri https://localhost:5001
When no HTTP version is specified explicitly HTTP/3 is used by default.
For available commands and options run the tool with the --help switch:
chttp --help
Currently supported commands and options:
Description:
Send HTTP request
Usage:
CHttp [command] [options]
Options:
-v, --http-version <1.0|1.1|2|3> The version of the HTTP request: 1.0, 1.1, 2, 3 [default: 3]
-m, --method HTTP Method [default: GET]
<CONNECT|DELETE|GET|HEAD|OPTIONS|POST|PUT|TRACE>
-h, --header <header> Headers Key-Value pairs separated by ':'. For example
--header="key:myvalue" []
-t, --timeout <timeout> Timeout in seconds. [default: 30]
--no-redirects Disables following redirects on requests [default: False]
--no-cert-validation, --no-certificate-validation Disables certificate validation [default: False]
-l, --log <Normal|Quiet|Verbose> Level of logging details. [default: Verbose]
-o, --output <output> Output to file. []
--cookie-container <cookie-container> A file to share cookies among requests. []
-b, --body <body> Request body or a file path containing the request
-u, --uri <uri> (REQUIRED) The URL of the resource
--upload-throttle <upload-throttle> Specify HTTP level throttling in kbyte/sec when sending the
request []
-k, --kerberos-auth Use Kerberos Auth [default: False]
--version Show version information
-?, -h, --help Show help and usage information
Commands:
forms Forms request
perf Performance Measure
diff Compares to performance measurement files
The tool supports HTTP/1.0, HTTP/1.1, HTTP/2 and HTTP/3. Specify 1.0, 1.1, 2, 3 values accordingly. For HTTP/3 requires support from the OS. On Windows this is available with recent Windows version and on Linux install the libmsquic package.
- Quiet: only a summary of the HTTP response is displayed.
- Normal: response headers and summary of the HTTP request is displayed.
- Verbose: response headers, content and summary of the HTTP request is displayed.
When multiple coherent requests would share a cookies, use a cookie containar. This cookie container will persist all cookies to a file, and updating this file after each request.
It is possible to throttle request content. Note, that this is only throttling HTTP level traffic. Specify the values in kbyte/sec.
Use simple string bodies or a filepath as the input for the request.
Note, that the current version of the tool does not allow sending JSON string in the body parameter. For more details see the linked issue. For sending JSON content, save the data to file and pass the filename as the argument of the body parameter.
Set the number of clients (-c) used to send the number of requestCount (-n) requests. The tool executes the test and writes out basic statistical information about the collected data.
chttp perf --help
Description:
Performance Measure
Usage:
CHttp perf [options]
Options:
-n, --requestCount <requestCount> Number of total requests sent. [default: 100]
-b, --body <body> Request body or a file path containing the request
-c, --clients <clients> Number of parallel clients. [default: 20]
-u, --uri <uri> (REQUIRED) The URL of the resource
--metrics <metrics> When Application Insights connection string is set, it pushes
performance metrics data. []
-v, --http-version <1.0|1.1|2|3> The version of the HTTP request: 1.0, 1.1, 2, 3 [default: 3]
-m, --method HTTP Method [default: GET]
<CONNECT|DELETE|GET|HEAD|OPTIONS|POST|PUT|TRACE>
-h, --header <header> Headers Key-Value pairs separated by ':'. For example
--header="key:myvalue" []
-t, --timeout <timeout> Timeout in seconds. [default: 30]
--no-redirects Disables following redirects on requests [default: False]
--no-cert-validation, --no-certificate-validation Disables certificate validation [default: False]
-l, --log <Normal|Quiet|Verbose> Level of logging details. [default: Verbose]
-o, --output <output> Output to file. []
--cookie-container <cookie-container> A file to share cookies among requests. []
-?, -h, --help Show help and usage information
Performance measurements yields results such as:
RequestCount: 100, Clients: 10, Connections: 10
| Mean: 322,698 us |
| StdDev: 80,236 us |
| Error: 8,024 us |
| Median: 310,700 us |
| Min: 198,700 us |
| Max: 652,700 us |
| 95th: 473,700 us |
| Throughput: 0.000 B/s |
| Req/Sec: 2,82E+04 |
------------------------------------------------------------------------
244,100 us #########
289,500 us ###################
334,900 us ###################
380,300 us ##############
425,700 us ####
471,100 us ##
516,500 us ##
561,900 us #
607,300 us
652,700 us #
------------------------------------------------------------------------
HTTP status codes:
1xx: 0, 2xx: 100, 3xx: 0, 4xx: 0, 5xx: 0, Other: 0
------------------------------------------------------------------------
The top section details standard statistical values. The middle section draws a distribution of the requests. The distribution is only rendered for performance measurements with at least 100 requests. The last section shows the response HTTP status codes of these results.
When results are persisted using the -o or --output parameters, the diff command can be used to compare results.
Description:
Compares to performance measurement files
Usage:
CHttp diff [options]
Options:
--files <files> List of 2 files to be compared. []
Such as:
diff --files session0.json --files session1.json
The command show the results of session0.json as the base and the results of session1.json as the difference to the base.
RequestCount: 100, Clients: 10
| Mean: 183.910 ms -1.847 ms |
| StdDev: 181.972 ms +6.523 ms |
| Error: 18.197 ms +652.321 us |
| Median: 114.454 ms +1.057 ms |
| Min: 102.744 ms -575.000 us |
| Max: 815.822 ms +19.062 ms |
| 95th: 735.611 ms +7.440 ms |
| Throughput: 114.016 KB/s +3.903 KB/s |
| Req/Sec: 52.3 +0.826 |
------------------------------------------------------------------------------------------------------------------------
175.441 ms ====================================================
248.712 ms ==
321.983 ms
395.255 ms
468.526 ms
541.798 ms
615.069 ms =+
688.341 ms #
761.612 ms =
834.883 ms ===
------------------------------------------------------------------------------------------------------------------------
HTTP status codes:
1xx: 0 +0, 2xx: 100 +0, 3xx: 0 +0, 4xx: 0 +0, 5xx: 0 +0, Other: 0 +0
------------------------------------------------------------------------------------------------------------------------
The distribution section uses = sign to indicate that both sessions have results in a given bucket; # where the base session have more results and + where the comparison sesoion has more results in the bucket.
dotnet tool install -g LaDeak.CHttpExec --prerelease
CHttpExec is tool that execute HTTP queries and HTTP performance measurements from a .chttp file. This can be useful to execute it on CI server or on a test infrastructure. When using the tool make sure that both the test target and the test executor machines are consistent in the available resources term, and no other concurrent apps share these resources during the performance measuremnt. This includes the network between the target server and the test executor.
Some automation infrastructure providers (such as GitHub Actions) might provide inconsistent amount of resources for different jobs.
Invoke the CLI as:
chttpexec Http3Tools\tests\test.chttp
A performance request can be attributed with # @assert expressions:
# @name validation
# @clientsCount 10
# @requestCount 100
# @assert mean < 1s stddev < 0.5s requestSec >= 0 throughput > 0 successStatus == 100
GET https://{{baseUrl}}/echo HTTP/2
{{nextRequest}}
Possible parameters to assert:
- mean
- stddev
- error
- median
- min
- max
- throughput
- requestsec
- percentile95th
- successStatus (the number of successful responses)
Values measured in time can be quantified by s ms us or ns. Values without a quantifier are processed as seconds.
Violations are reported as:
ASSERTION VIOLATION
error: Mean is not < 1.000ns
error: StdDev is not < 0.001ns
MeasurementsSession allows to apply the same logic used by the CHttp CLI, dotnet tool and VS Extension in C# projects.
var session = new MeasurementsSession("https://localhost:5001");
for (int i = 0; i < 10; i++)
{
session.StartMeasurement();
// Execute measured code...
session.EndMeasurement(HttpStatusCode.BadRequest);
}
await session.PrintStatsAsync();- Use
DiffAsyncto create diffs from files - this allows to compare files programatically from sources of measurements. - Use
SaveAsyncto save a measurement session into a file. - Use
PrintStatsAsyncto print the results on the console. GetSessionandDiffmethods allow to compare measurement sessions in memory.
Run the following commands to publish the native dependencis of the VS Code Extension and to copy them to the extension's dependencies:
dotnet publish src/CHttpExtension -r win-x64
cp ./src/CHttpExtension/bin/Release/net9.0/win-x64/publish/* ./src/VSCodeExt/src/chttp-win-x64
- Clear npm cache
npm cache clean --force - Remove yeoman
npm install -g yo generator-code - Uninstall
npm install -g @vscode/vsce - Uninstall npm