I love how #Chainguard Images continues to patch security issues before there is even a #CVE issued.

- #CVE-2023-38403 (#iperf3): Patched same day as announcement (8 days before CVE)
- #CVE-2023-3446 (#openssl): Patched same day as announcement
- #CVE-2023-38408 (#openssh): Patched same day as announcement

That's the beauty of automation and testing.

Hm, this can actually be the #MacOS #Ventura 13.3.1 update, seems to have made networking quite funky. Between reboots right after another, I've seen:
* normal download / broke upload;
* half-speed download / normal upload;
* normal speed but retry-heavy download / normal-ish upload

On WiFi, measured with #iperf3 to another server on the LAN. A bit weird one,, have to keep an eye out for it...

Continuation of my network problems between #Windows 10 and certain remote network behind a #Mikrotik. Funny fact - Linux client from same local network works fine. And today I've found out that #iperf3 running in parallel mode achieves proper speed as well - though streams are slow as before, but the sum is good. Now I wonder WTF is going on with single streams. And I'm talking about up to 60 Mbps/s, not Gigabits of bandwidth.

Continuation of my network problems between #Windows 10 and certain remote network behind a #Mikrotik. Funny fact - Linux client from same local network works fine. And today I've found out that #iperf3 running in parallel mode achieves proper speed as well - though streams are slow as before, but the sum is good. Now I wonder WTF is going on with single streams. And I'm talking about up to 60 Mbps/s, not Gigabits of bandwidth.

#Network gurus, any idea why would (#Windows 10?) throttle upload network bandwidth to a specific (#mikrotik) router? computer->mikrotik->5G->FTTH->mikrotik->server. It always throttles it to ~2Mb/s (it should be at least 10Mb/s). From all the combinations I tested, it seems happening only with this computer to that router. From Mac and phones it works as expected. Testing using #iperf3.

Testing #iperf3 for #sctp --nstreams is definitely not very usable/functional at all. Looks like it uses the client buffer to cue up 64k of data... and that's it

Automatisation tests de réseau avec #iperf3 : https://www.redhat.com/sysadmin/network-testing-iperf3