Re: Solving the syslogd problem

[mlelstv%serpens.de@localhost (Michael van Elst)]

christos%astron.com@localhost (Christos Zoulas) writes:

>Or have 2 syslogd binaries built from the same sources. One with only
>localhost support in /sbin which starts up early during boot and another
>which replaces it later when /usr is mounted. Having a split /usr makes
>little sense today though as joerg mentioned, even in the space-constrained
>systems.


If the "syslogd problem" isn't just a "dhcpcd logging problem", removing
the /usr split or the proposed more radical agendas won't be enough.
You also need to log somehwere to and the "/var split" is then the next
thing that must go away.

An early logging daemon that buffers messages in memory until they can be
written out solves that part too (it's a systemd feature!).

The traditional approach implemented by syslog() is to send output to console
until syslogd is available (with LOG_CONS). We could just treat this output
like kernel messages and solve the "syslogd restart gap problem" with it too.

-- 
-- 
                                Michael van Elst
Internet: mlelstv%serpens.de@localhost
                                "A potential Snark may lurk in every tree."