[Nottingham] systemd now puts "mount" into the magical "PID 1"

[VM]

On 23 August 2016 14:47:24 BST, VM <vadim at mankevich.co.uk> wrote:
>I fell in love with Systemd when it did just that. I started to hate it
>when it became so complicated that fault searching now sometimes is
>nearly impossible. Unless you've read changelogs to 231 versions and
>actually understood what they mean. BTW, some functionality is only
>documented in the code and changelogs... And then distros and file
>system developers introduce their own fixes and workarounds just to
>accommodate this rapidly expanding system that fixes what even wasn't
>broken.
>I think it makes my systems more fragile, prone to subtle
>incompatibilities and less controllable. My recent incident with
>systemd on Debian proves the point and made me seriously looking in the
>direction of Devuan, OpenRC on Arch/Manjaro, Void Linux with runit and
>even Gentoo.
>My "favourite" part of systemd is hostnamectl. It's a service whose
>only purpose is to read or write one word in /etc/hostname (and
>/etc/machine-info). As if "cat" is not suitable enough. :(
>Yes, it does a good job monitoring services. That is all I want from
>it. Not a DNS resolver that may have completely different view of the
>world to the rest of the system. Not any other BS. And then it should
>be PID 2 after something small and beautiful like
>https://github.com/siblynx/uinit/blob/master/init.c 
>
>I have one linux container that sends dhcp requests despite being
>configured with static network settings and without any dhcp clients
>installed. If I find out it's systemd that will be an end to my
>relationship with systemd. :)
>
>Hat off for distro maintainers who have to keep their systems alive
>regardless of what next thing systemd would consume or change. But then
>again they chose systemd in the first place... Karma :)
>
>> On 23 Aug 2016, at 14:01, Jason Irwin via Nottingham
><nottingham at mailman.lug.org.uk> wrote:
>> 
>>> On 23/08/16 12:03, Neal Ponton via Nottingham wrote:
>>> I don't think I've managed to create a 'simple' systemd unit yet
>without
>>> it coming up as having problems when running 'systemctl status'.
>> If you look on the NLUG site, you'll see a quick example of using
>systemd to create an ssh tunnel.
>>
>http://nlug.ml1.co.uk/2016/03/resilient-ssh-with-autossh-andor-systemd/5148
>> 
>> I also found using systemd to ensure the VPN client runs at the right
>time to be a snap.
>> 
>> For some other things...I've just stuck to a crontab; they didn't
>need to fuss with system events.
>> 
>> I really like some of the stuff systemd makes easy "This wants that,
>provides other". Makes much more sense to my tiny mind that have
>fragments scattered all over in rc.local, if-up.d etc. (and potentially
>in different places depending on the SysAdmin's style). But then I
>probably don't know enough to understand all the pitfalls.
>> 
>> The PID 1 thing though does worry me more and more though. This is a
>succinct explanation of the issues.
>> http://ewontfix.com/14/
>> 
>> Anyone have a rebuttal from the systemd-crew on why PID 1 isn't an
>issue?
>> 
>> -- 
>> ╔═════════════╦══════════════════════════════════════════╗
>> ║ Jason Irwin ║ OpenPGP (GPG/PGP) Public Key: 0xD0C592B1 ║
>> ║             ║ Import from hkp://pgp.mit.edu            ║
>> ╚═════════════╩══════════════════════════════════════════╝
>> 
>> _______________________________________________
>> Nottingham mailing list
>> Nottingham at mailman.lug.org.uk
>> https://mailman.lug.org.uk/mailman/listinfo/nottingham