SCIENTIFIC-LINUX-USERS Archives

June 2015

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
ToddAndMargo <[log in to unmask]>
Reply To:
ToddAndMargo <[log in to unmask]>
Date:
Sat, 13 Jun 2015 22:43:15 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (82 lines)
On 06/13/2015 09:26 PM, Steven Haigh wrote:
>> On 14/06/2015 1:35 PM, Jamie Duncan wrote:
>>> On Sat, Jun 13, 2015, 11:32 PM ToddAndMargo <[log in to unmask]
>>> <mailto:[log in to unmask]>> wrote:
>>>
>>>      On 06/13/2015 08:25 PM, Jamie Duncan wrote:
>>>      > You define what each service needs or wants for it to be able to
>>>      start.
>>>      > Look at the .service files
>>>      >
>>>      >
>>>      > On Sat, Jun 13, 2015, 11:23 PM ToddAndMargo <[log in to unmask]
>>>      <mailto:[log in to unmask]>
>>>      > <mailto:[log in to unmask] <mailto:[log in to unmask]>>>
> wrote:
>>>      >
>>>      >     Hi All,
>>>      >
>>>      >     /etc/rc/d/rc5.d with its numbered start points are being
>>>      >     phased out.  In SL7, systemctl, what determines when things
>>>      >     start?
>>>      >
>>>      >     Many thanks,
>>>      >     -T
>>>      >
>>>
>>>      # find /etc/systemd/system/ -iname \*.service | grep -i firewall
>>>      /etc/systemd/system/dbus-org.fedoraproject.FirewallD1.service
>>>      /etc/systemd/system/basic.target.wants/firewalld.service
>>>
>>>
>>>      Is this it (firewalld.service)?
>>>           [Unit]
>>>           Description=firewalld - dynamic firewall daemon
>>>           Before=network.target
>>>           Before=libvirtd.service
>>>           Before=NetworkManager.service
>>>           Conflicts=iptables.service ip6tables.service ebtables.service
>>>
>>>      So you tell it before what you want?  Seems a bit confusing.
>>>      Am I missing something?
>>
>> There are many options to allow it to be parallel. It's well documented
>> upstream and for RHEL. Time to hit the books!
>
> What he's really saying is that there is no set order. There is no
> reliable order. There is no pre-determined order. What you do get is a
> setting that says "I will load service X before Y" - which it has to
> work out on every boot.
>
> It may or may not be the same order every time you boot - and depending
> on what is set where, it may even start services that are not
> specifically enabled if they are listed as dependencies.
>
> One good example is upowerd. Even if you have this disabled, it will
> still get started by other services in the boot process.
>
> Now you can argue until the cows come home as to if this is good or not
> - and I'll leave that to an academic argument.
>
> Before and After in the unit file can be abused - and you could end up
> very easily in a logic loop that is not possible. I'm not sure how it
> tries to deal with these cases - or if it just throws its hands up in
> the air and complains (verbally or silently).
>
> Consider the following:
> a.service:
> 	[Unit]
> 	Before b.service
>
> b.service:
> 	[Unit]
> 	Before a.service
>
> Now generally, if you only use distro supplied packages, this should be
> fine - however it does make things more difficult for the packager.
>
> --
> Steven Haigh

Thank you!

ATOM RSS1 RSS2