Austin S. Hemmelgarn
Austin S. Hemmelgarn
> Relying on CMake's patching, This is functionality that has existed in CMake for years and is generally considered perfectly stable. And all that it is internally is a `add_custom_command()`...
Rebased to resolve merge conflicts.
> It is true that the socket is in the wrong directory. I think that /run or /var/run matches the semantics better and we can create e.g. /var/run/netdata/ at install-time...
@mfundul It works for both root and whatever user `netdata` is running as. I think in some cases it might also work for people in the `netdata` group, but I'm...
> > @mfundul It works for both root and whatever user `netdata` is running as. I think in some cases it might also work for people in the `netdata` group,...
I’m planning to pick this up some time in the near future as it’s blocking a couple of other issues and should (in theory) be an easy fix.
Yes, we still need this, because nothing has changed and the issues I outlined with the current approach in the original post are _still issues_. I’m not sure if anybody...
Just a thought, but we probably want the auto-generated vnode UUID to be stable for a given combination of Netdata agent ID and vnode name so that the association between...
I’m not 100% certain if this is still an issue at the moment or not.
The only resolution for this that gives us 100% consistency across all install types and installation conditions is to just _not_ fetch the config on any install type, and expand...