[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Heads up: persistent journal has been enabled in systemd



Hi,

On Thu, Feb 06, 2020 at 07:24:28PM +0100, Ansgar wrote:

[popcon is not indicative of install base]

> That's likely given some libreoffice packages (unlikely to be installed
> on servers) are at ~40% popcon and I would expect significantly more
> server installations than desktop ones.

> So popcon might overestimate sysvinit usage and it might in reality be
> lower over the total installation base.

Also quite possible.

That's my point: we do not have useful data on deployments because desktops
are more likely to report in, container builders are more likely to
repeatedly download the same package in rapid succession, container users
will often not download an init system from the mirror network at all
because it is in the base images, datacenter operators will certainly run
their own mirrors, embedded systems are deployed via dd'ing images,
company-wide FAI rollouts almost certainly disable popcon, containers use
the service as the entry point and skip the init system, dependencies might
pull in packages that go unused, ...

That is what I'm saying: these numbers have more noise than signal in them,
especially for the init systems. The numbers are very useful for comparing
packages within a suite, e.g. GNOME Terminal vs GNOME Minesweeper, which is
why we use them to distribute packages over the installation media, but I
wouldn't even trust them to get the ratio of GNOME vs KDE installations
right.

If we wanted to get more accurate statistics, we'd have to at least
distinguish between bare metal, VMs, and containers on one axis, and
manual installation by the user, default installation by d-i or automated
installation through a configuration management tool, and then we'd still
have to correct for the fact that packages installed by default will not
usually be manually installed. We'd have slightly more accurate numbers
that would still give anyone without a relevant doctorate pretty much
nothing to work with, and it would be a lot of effort, so let's not.

tl;dr: statistics is hard, please do not base decisions on popcon or
downloads.

   Simon


Reply to: