Ensure an iperf3 of 1-128 streams will not result into a crash, by a simple e2e stream test among containers

[fgeorgatos]

What would you like to see added/updated/deleted in our end-to-end tests?

Let's add a scaling iperf3 test.

Back in the day, it became possible to hit and uncover a linux kernel bug which was triggered from user-space:

• reproducible kernel panic w. 4.19.0 & parallel iperf threads P>8, weave/2.5.* weaveworks/weave#3684

In all fairness, this was not weave-fastdp-specific (which was an aggravating factor) but rather linux kernel-version-specific.

A beauty of this test is its simplicity: it just has to be run in user space, across two containers over the CNI as follows:

• iperf3 -s
• echo 1 2 4 8 16 32 64 128|xargs -n1 iperf3 -c <server_cni_ip> -P

Would this be of interest? Is it covered as a test already in some other way? Does it belong elsewhere perhaps?

Why is this needed?

Ensure that before shipping a release this test has been run successfully at least once, so it can be used as a reference for validation of deployment environments (i.e. in downstream installations).