How to use the iperf library to obtain bandwidth information in real-time at intervals?

[QingshuangZhu]

I compiled iperf into a library and then used it in my code. I want to obtain the bandwidth information once every interval.
What should I do? Thank you.

[davidBar-On]

As far as I am aware there is no such capability in iperf3. One iperf_run_client(() is called, there is practically no interaction with the library, except for the possibility to view the output (stdout or the iperf_set_test_logfile(() file).

[QingshuangZhu]

Thank you very much for your answer. I took a brief look at the iperf3 source code and it indeed does not support this feature.

Currently, I am doing it like this: first, I set the JSON streaming output with iperf_set_test_json_output(test, 1) and iperf_set_test_json_stream(test, 1); And I set logfile and outfile through iperf_set_test_logfile("./iperf3.log"); then I execute iperf_run_client(test) in one thread, and another thread waits for 2 seconds before reading the JSON data from iperf3.log and parsing out the bandwidth information.

I look forward to iperf3 supporting this feature through a callback function. Thank you again.

[davidBar-On]

Thanks for the detailed explanation. Not sure how easy/difficult it is to implement that, but Just to make sure. Are you looking for a callback at the end of the test, so there will be no need to wait the 2 seconds, something like the function bellow that will set JSON file output and will define the callback at the end of the test, something like:

void end_of_test_callback(struct iperf_test *test);
int iperf_set_test_json_end_of_test_callback(
    struct iperf_test,
    (*json_callback)(struct iperf_test *),
    char *logfile_name
);

Or, something else?

[QingshuangZhu]

Thank you very much for your reply. I hope to obtain bandwidth information in real-time at each time interval, rather than at the end of the test. The 2 seconds wait is to ensure that the test data is written into the logfile file. I think an interval event callback function should be added, but I feel it might not be easy to implement (I am not familiar with the iperf3 source code).