Re: Four people troll and systemd FUD
On 04/03/14 00:40, Stephan Seitz wrote:
> On Mon, Mar 03, 2014 at 12:52:40PM +1100, Scott Ferguson wrote:
>> own personal computers my sentiments are similar. However my business
>> purposes involve meeting SLAs so reboots once or twice a year can cost a
>> lot of money - so in those circumstances a few minutes makes a lot of
>> difference. Perhaps that's not something you care about - or it's just
>
> Sorry, I don’t buy this.
That's fine, you don't have to.
> If your systems are virtual machines then a
> reboot is already fast.
How fast? Under 6 seconds?
Perhaps you confuse web hosting (for non-critical services) the only
business application? (I've read your posts before - so I seriously
doubt that).
> Filesystem checks may delay the reboot, or
> applications that need minutes to stop or start, but systemd doesn’t
> help here either.
Actually it can (and I'm sure you can do your own homework), but I'm not
defending systemd - only Debian's right to choose it (for Linux).
A. I don't have an opinion on systemd - yet[*1]. I've read all the
arguments made by those in a position to know what they're talking
about, but won't hold an opinion until I've seen the results of a lot of
testing.
B. Just because NOC and/or mid-range don't have their act together
doesn't mean back-end gets a holiday. All the components within a system
need to be optimised.
[*1] from a business point of view any change is costly, and the gains
from better service control and other features including faster reboots
will have to be measured against the inherent cost of implementing the
changes. And no system is perfect - from an engineering point of view
ditching a system that 99% works in the hope that a new-from-scratch
system will work 100% is um, optimistic - but that needs to be balanced
against the reality that tacking new features onto something that 99%
works (but lacked desired features) has very little chance of long-term
success (maintenance and documentation quickly exceed any benefits).
>
> If your systems are real server hardware then your reboot is mainly
> delayed by the BIOS.
OK - you can stop there, come back when you've had your morning coffee
and had a think about what all the computers around (like your phone)
are not using. :)
You're also missing the point about "all the components needing
optimising - I can only presume you work in a non-competitive area,
which history shows doesn't remain that way for long.
Hint:- as far as I'm concerned systemd is in the future (maybe). Rarely,
very rarely, in critical application support is Testing or Unstable put
into production - by the time that happens it'll be on hardware that
won't have a BIOS (that's the itch that the funding for Coreboot
development scratches.)
> Here any server (blade or normal) takes much longer
> from BIOS to bootloader than from bootloader to login prompt.
>
>> Fast booting was not the sole criteria for which it was selected by
>> Debian for the *Linux* kernel.
>
> True, but I don’t need any of the new features (never had any problems
> with sysvinit). So why should I change?
Agreed - though I don't know your use case.
Why should you care if others don't agree?
>
> Shade and sweet water!
>
> Stephan
>
Kind regards
Reply to: