> I'm running a pretty conventional setup with a server providing > nat/dhcp/dns/nfs/yp to about 15 workstations. These nfs mount /home
> via /etc/rc.local from where they find another script that nfs
> mounts /usr and does some trivial local house keeping.
> My reluctance to use systemd is mainly that the documentation seems to
> be fragmented and lacking a decent overview, and I'm never keen on
> learning a system by hacking it about until it Seems To Work. On the
> other hand I know that my existing configuration files for sysv will
> Just Work, or require minimal tweezing at most.
Test, test, test, test ...
Make clone/copy of system to run on a disconnected test network
with a clone of workstation. Test the upgrade to jessie.
Test all the services booted under sysvinit and systemd init...
Then consider test as a separate address / computer name on main
network if appropriate, change client over to connect to new
hostname and check?
Don't worry about it unduly, just ... manage, document process
you took in test-upgrade, be careful =)...
> However, it doesn't look as if systemd is going away any time soon, and
> I suspect that hybrid solutions may be overly complex for my needs. Treat debian as a hybrid already =)....
I wouldn't be concerned by using sysvinit scripts that work =).
Main thing is to KNOW what you are relying on, not miss
some important aspect to the system that creates a disruptive
regression... test test test =).
This message was posted to the following mailing lists: