Curl Command Send Asynchronous Request

I using this curl request for sending data to API :

curl --location --request POST 'MY_IP' \
--header 'Content-Type: text/plain' \
--data-raw ' [{
    "event_id": "123",
}]    
'

I want to check the API performance by sending request asynchronously and measure the time take I tried to use xargs -P 10 which say send 10 request parallel but not able to do so ? Any help and how I can measure time ?

My Code :

#!/bin/bash

Body='{
    "event_id": "12",
    "metric_name": "API",
    "value" : "1",
    "dimensions": {  
    },
    "timestamp_ms": 1615552313
}';

seq 1 2 | xargs -n1 -P10 curl --location --request POST '10.33.137.98:8080' --header 'Content-Type: text/plain' --data-raw "$Body"

But how to calculate time like P95 , P90 etc ?

I also couldn't get xargs -P working when I needed to do something similar, and I ended up using &.

From man bash:

If a command is terminated by the control operator &, the shell executes the command in the background in a subshell. The shell does not wait for the command to finish, and the return status is 0. These are referred to as asynchronous commands.

So something like this, maybe?

#!/bin/bash

Body='{
    "event_id": "12",
    "metric_name": "API",
    "value" : "1",
    "dimensions": {  
    },
    "timestamp_ms": 1615552313
}';

function run() {
    for i in $(seq 1 10); do
        echo "running task $i"
        curl --location --request POST '10.33.137.98:8080' --header 'Content-Type: text/plain' --data-raw "$Body" &
    done 

    wait
}

time run

Explanations:

• time is used to measure total execution time, see man time
• wait ensures that all the background jobs finish before continuing, otherwise the time would just measure how long to launch all the requests, not how it took to get a response
• parallel might also work instead of &

To calculate statistics, you could output the time of each request, using a second time call in the for loop, and then process it using your favourite analysis tools.

Make a bash function. Call it from GNU Parallel. Use --joblog to see timing.

doit() {
    curl --location --request POST 'MY_IP' \
    --header 'Content-Type: text/plain' \
    --data-raw ' [{
        "event_id": "123",
    }]    
    '
}
export -f doit
seq 10000 | parallel -j 10 --joblog my.log doit

Now look at my.log.

GNU Parallel has an overhead of 1-10 ms per job, so if your curls are really short, you could do more than one:

seq 10000 | parallel -j 10 --joblog my.log 'doit;doit;doit;doit;doit;doit;doit;doit;doit;doit'