systemd timing

[Xen]

Karl Auer schreef op 16-07-2017 17:01:

> In trying to make that simple thing happen, I was exposed to the full
> lunacy that is systemd.
> 
> I finally got it to work by making it dependent on a late-completion
> target ("WantedBy=multi-user.target"). I was completely unable to
> locate the correct moment to re-run it, so the fine-grained control
> that is supposed to be enabled by systemd was made a mockery of.

Well I guess you can do pre and post stuff (Before and After) wrt other 
services but that means you get a solution that is rather specific to 
other services existing, I guess...

In the end it works and then you wonder why ;-). Right.

Amazing that you got this far in any case. Hats off.

> The solution to my actual problem was to use static IPv6 addresses
> (which is what I needed anyway, I just got distracted by SLAAC
> addressing apparently not working).

I think SLAAC is not very useful myself, but yeah... as if it still 
matters what I'd say or what anyone would say ;-).

The only thing you can do is try to stay true to yourself in that sense.

> There are still several anomalies WRT IPv6 addressing; I think the
> presence of at least three systems, each of which seems to want to
> control elements of SLAAC, is a big part of the problem. DHCP, systemd
> and the "old" /etc/network/interfaces method.

Alright.

I was never bothered by IPv4 in the slightest, and then when there was 
some debate somewhere my 'opponent' said he liked IPv6 because it meant 
he could easier manage the 100 computers behind the firewall.

And I'm like, but if you are going to be doing things on that scale, 
surely a software could be developed to act as a proxy for that in an 
easier way?

IPv4 throws up hindrances, but they only really start to bug people when 
the scale goes up, and at that point you wonder why they can't invest 
time in finding another solution ;-).

Anyway, so much for my unsolicited opinion.

Regards.