Tristan

My bad... Seems like I picked the wrong version to deploy... I'll try again...

So, funny thing is: - I did deploy the wrong version earlier (I think, it's confusing) - I deployed the right one (with the patch) for sure now - The...

I’ll send those over

I can confirm that the bug is gone with your patch on top of master :+1:

Digging a bit more, this seems like it's also been reported a long time ago here (in 1.5) https://haproxy.formilux.narkive.com/WkdV7fGj/abstract-namespace-sockets-handling

Agreed on both points; as you'll see on the ML, I took a tunable approach as I'm sure some people manually tweaked their other programs already. As for changing the...

> So yes, I'd say changing the default behavior of the abns@ address family prefix is not possible. While I don’t think it’s mandatory, I don’t agree that it’s impossible...

Here's the dashboard I was mainly looking at We have 4 timeranges: - 3.0-dev1 - 3.0-dev4 without tuning - 3.0-dev4 with only the frontend buffers tuning - 3.0-dev4 with both...

I don't know how useful it is but here's the average sysctls for one of our nodes (they're not all perfectly uniform everywhere, but this should be mostly consistent). Note...

> It's strange that the sysctl didn't list net.core, I was particularly interested in this one, more specifically {r,w}mem_{default,max}. My bad for that. Our nodes run with LXC; sysctls are...