Austin S. Hemmelgarn

I’m honestly not actually sure why the script is still around, and am kind of surprised that it’s been working at all over the past few months. The current recommended...

I think we need to just reevaluate the entire contrib directory. I’m not sure the rhel stuff is useful anymore, and the debian directory should ideally be in the root...

_Most_ of the work required for this should be done now, at least for the CPU architectures we provide static builds for (currently 64-bit x86, ARMv7a, AArch64, and POWER8+). I...

To clarify on the point about bash still being required, the last major bit is _probably_ the stuff being removed by https://github.com/netdata/netdata/pull/14829. Once that gets merged I plan to make...

This _should_ be a reasonably safe change, as zlib is available pretty much everywhere (the only people likely to be affected are those building custom systems from the ground up....

Current plan is to merge this tomorrow if possible (waiting one day so we have at least one nighly after the new release without any potentially breaking changes).

Just an observation, but we may want to consider allowing one sender _per upstream parent_. Having a single thread covering all the children does make sense, but on large scale...

> How about using io_uring on supported systems to minimise the connection costs and reduce context switches? `readfile()`, once that’s finally widely available, is probably going to be a much...

> > > How about using io_uring on supported systems to minimise the connection costs and reduce context switches? > > > > > > `readfile()`, once that’s finally widely...

Repeating here what was discussed about this on Slack: - The logical split is to have the CO-RE code bundled with the eBPF plugin, and have the ‘legacy’ code for...