visualise openrc initscript start order and dependency tree

classic Classic list List threaded Threaded
15 messages Options
Reply | Threaded
Open this post in threaded view
|

visualise openrc initscript start order and dependency tree

William Kenworthy
I have run into some problems creating openrc initscripts for moosefs -
is there something that will display the start order/dependency tree? 
Text or graphical doesn't matter.

Bill K.




Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

Daniel Frey-6
On 2019-11-07 04:37, Bill Kenworthy wrote:
> I have run into some problems creating openrc initscripts for moosefs -
> is there something that will display the start order/dependency tree?
> Text or graphical doesn't matter.
>
> Bill K.
>
>
>
>

rc-status does try to resolve dependencies and list them in order, try
`rc-status -a` - this will list all scripts (including ones not in startup.)

Dan

Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

J. Roeleveld
On Thursday, November 7, 2019 4:06:10 PM CET Daniel Frey wrote:

> On 2019-11-07 04:37, Bill Kenworthy wrote:
> > I have run into some problems creating openrc initscripts for moosefs -
> > is there something that will display the start order/dependency tree?
> > Text or graphical doesn't matter.
> >
> > Bill K.
>
> rc-status does try to resolve dependencies and list them in order, try
> `rc-status -a` - this will list all scripts (including ones not in startup.)
>
> Dan

Actually, it doesn't show what is depending on what.

I had a similar issue and ended up checking every init-script, conf.d file and
rc.conf entry and making a dependency-tree manually on a big white-board.

I haven't found a tool that does this automatically yet.

--
Joost




Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

J. Roeleveld
In reply to this post by William Kenworthy
On Thursday, November 7, 2019 1:37:21 PM CET Bill Kenworthy wrote:
> I have run into some problems creating openrc initscripts for moosefs -
> is there something that will display the start order/dependency tree?
> Text or graphical doesn't matter.
>
> Bill K.

The closest to this is:

/lib/rc/bin/rc-depend  xdm

(This will return ALL the services that need to be started before "xdm" can
start, according to the dependencies)
This can also be executed as a normal (NON-root) user.

To force a regen of the dependency-tree, you can issue:
"/lib/rc/bin/rc-depend --update" (This does require root)

--
Joost



Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

William Kenworthy
In reply to this post by J. Roeleveld
On 9/11/19 4:03 am, J. Roeleveld wrote:

> On Thursday, November 7, 2019 4:06:10 PM CET Daniel Frey wrote:
>> On 2019-11-07 04:37, Bill Kenworthy wrote:
>>> I have run into some problems creating openrc initscripts for moosefs -
>>> is there something that will display the start order/dependency tree?
>>> Text or graphical doesn't matter.
>>>
>>> Bill K.
>> rc-status does try to resolve dependencies and list them in order, try
>> `rc-status -a` - this will list all scripts (including ones not in startup.)
>>
>> Dan
> Actually, it doesn't show what is depending on what.
>
> I had a similar issue and ended up checking every init-script, conf.d file and
> rc.conf entry and making a dependency-tree manually on a big white-board.
>
> I haven't found a tool that does this automatically yet.
>
> --
> Joost
>
>
>
>
Thanks, I ended up working around it rather than solving i.


BillK




Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

Neil Bothwick
In reply to this post by J. Roeleveld
On Fri, 08 Nov 2019 21:03:13 +0100, J. Roeleveld wrote:

> I had a similar issue and ended up checking every init-script, conf.d
> file and rc.conf entry and making a dependency-tree manually on a big
> white-board.
>
> I haven't found a tool that does this automatically yet.

systemd's systemctl ;-)

I'll get my coat...


--
Neil Bothwick

Some cause happiness wherever they go. Others whenever they go.

attachment0 (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

J. Roeleveld
On 9 November 2019 11:42:38 CET, Neil Bothwick <[hidden email]> wrote:

>On Fri, 08 Nov 2019 21:03:13 +0100, J. Roeleveld wrote:
>
>> I had a similar issue and ended up checking every init-script, conf.d
>> file and rc.conf entry and making a dependency-tree manually on a big
>> white-board.
>>
>> I haven't found a tool that does this automatically yet.
>
>systemd's systemctl ;-)
>
>I'll get my coat...

Does this parse openrc scripts correctly?
--
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

Rich Freeman
On Sat, Nov 9, 2019 at 7:01 AM J. Roeleveld <[hidden email]> wrote:

>
> On 9 November 2019 11:42:38 CET, Neil Bothwick <[hidden email]> wrote:
> >On Fri, 08 Nov 2019 21:03:13 +0100, J. Roeleveld wrote:
> >
> >> I had a similar issue and ended up checking every init-script, conf.d
> >> file and rc.conf entry and making a dependency-tree manually on a big
> >> white-board.
> >>
> >> I haven't found a tool that does this automatically yet.
> >
> >systemd's systemctl ;-)
> >
> >I'll get my coat...
>
> Does this parse openrc scripts correctly?

Only if somebody has created a generator for openrc, which I doubt.
It was obviously a semi-trollish comment.

That said, if somebody is looking to write a utility and wants to see
what else is out there, systemd has a couple of useful utilities that
work along these lines.

For example, a short snippet out of systemctl list-dependencies:
●   ├─remote-fs.target
●   │ ├─mnt-lizardfs.mount
●   │ ├─var-lib-machines.mount
●   │ └─nfs-client.target
●   │   ├─auth-rpcgss-module.service
●   │   ├─rpc-statd-notify.service
●   │   └─remote-fs-pre.target

(The full output is obviously long on any complex host, but this gives
an idea.  A target is basically a virtual service - similar to a
virtual package like plasma-meta.)

There are also some tools in systemd-analyze that give you
bootchart-like capabilities based on log scanning.  It can figure out
the critical path to getting to your target runlevel and what services
slowed that down the most (though on a really busy host you need to
bear in mind that other stuff was launching in parallel so the times
given aren't what you'd get if nothing else was running.)

Oh, and while bootchart doesn't really show dependencies (at least I
don't think it does), if you haven't checked it out you might still
find it useful.  I believe that does work with openrc, and it at least
helps you see where the system is spending its time at boot.

--
Rich

Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

antlists
On 09/11/19 19:51, Rich Freeman wrote:

> On Sat, Nov 9, 2019 at 7:01 AM J. Roeleveld <[hidden email]> wrote:
>>
>> On 9 November 2019 11:42:38 CET, Neil Bothwick <[hidden email]> wrote:
>>> On Fri, 08 Nov 2019 21:03:13 +0100, J. Roeleveld wrote:
>>>
>>>> I had a similar issue and ended up checking every init-script, conf.d
>>>> file and rc.conf entry and making a dependency-tree manually on a big
>>>> white-board.
>>>>
>>>> I haven't found a tool that does this automatically yet.
>>>
>>> systemd's systemctl ;-)
>>>
>>> I'll get my coat...
>>
>> Does this parse openrc scripts correctly?
>
> Only if somebody has created a generator for openrc, which I doubt.
> It was obviously a semi-trollish comment.
>
Now that's harsh! Although yes I'm sure he was tweaking tails - hence
the "tongue in cheek" smiley.

Fact is, there are a lot of people out there who hate systemd because
it's been successful, and it's been successful because it sticks to the
nix philosophy of "do one thing, and do it well".

Cheers,
Wol


Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

Rich Freeman
On Mon, Nov 11, 2019 at 5:38 AM Wols Lists <[hidden email]> wrote:
>
> On 09/11/19 19:51, Rich Freeman wrote:
> >
> > Only if somebody has created a generator for openrc, which I doubt.
> > It was obviously a semi-trollish comment.
> >
> Now that's harsh! Although yes I'm sure he was tweaking tails - hence
> the "tongue in cheek" smiley.

I didn't intend to suggest that I thought it was mean-spirited.  Just
stirring the pot.

>
> Fact is, there are a lot of people out there who hate systemd because
> it's been successful, and it's been successful because it sticks to the
> nix philosophy of "do one thing, and do it well".
>

Now, THAT is a semi-trollish comment if I ever saw one.  :)

That said, you could argue that the individual components of systemd
do generally do one thing well.  I think the criticism is more in the
packaging, and that the components mostly don't interchange with
anything non-systemd.  Though as we can see from eudev/elogind and so
on that isn't strictly the case.

I sometimes describe systemd as the anti-busybox.

But, I don't want to derail the thread entirely...

--
Rich

Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

Mick-10
On Monday, 11 November 2019 11:54:31 GMT Rich Freeman wrote:
> On Mon, Nov 11, 2019 at 5:38 AM Wols Lists <[hidden email]> wrote:
>
> > Fact is, there are a lot of people out there who hate systemd because
> > it's been successful, and it's been successful because it sticks to the
> > nix philosophy of "do one thing, and do it well".
>
> Now, THAT is a semi-trollish comment if I ever saw one.  :)

Well, the major criticism *against* systemd has been that it has been designed
in an orthogonal direction to the *nix philosophy.  It tried from inception to
do many things, building a monolithic stack primarily to facilitate quick and
easy spinning of linux deployment in cloud technologies.


> That said, you could argue that the individual components of systemd
> do generally do one thing well.  I think the criticism is more in the
> packaging, and that the components mostly don't interchange with
> anything non-systemd.  Though as we can see from eudev/elogind and so
> on that isn't strictly the case.
>
> I sometimes describe systemd as the anti-busybox.

Well, some systemd components can be taken as single applications and used
separately from the whole systemd stack, that much is true.  However, (some)
systemd devs are known for for being disrespectful towards the rest of the
Linux ecosystem and making architectural decisions which break
interoperability.  systemd has been gradually taking over more and more
functions/services which reminds me of the old emacs joke:

 "... emacs is a fine operating system, in need of a good editor"


> But, I don't want to derail the thread entirely...

Sorry, I couldn't resist contributing!  :-)

--
Regards,

Mick

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

antlists
In reply to this post by Rich Freeman
On 11/11/19 11:54, Rich Freeman wrote:

> On Mon, Nov 11, 2019 at 5:38 AM Wols Lists <[hidden email]> wrote:
>>
>> On 09/11/19 19:51, Rich Freeman wrote:
>>>
>>> Only if somebody has created a generator for openrc, which I doubt.
>>> It was obviously a semi-trollish comment.
>>>
>> Now that's harsh! Although yes I'm sure he was tweaking tails - hence
>> the "tongue in cheek" smiley.
>
> I didn't intend to suggest that I thought it was mean-spirited.  Just
> stirring the pot.
>
Fair enough. Trouble is, you can't be too careful with what you say on
the internet - it's too easily taken out of context or just plain
mis-understood.

>>
>> Fact is, there are a lot of people out there who hate systemd because
>> it's been successful, and it's been successful because it sticks to the
>> nix philosophy of "do one thing, and do it well".
>>
>
> Now, THAT is a semi-trollish comment if I ever saw one.  :)
>
> That said, you could argue that the individual components of systemd
> do generally do one thing well.  I think the criticism is more in the
> packaging, and that the components mostly don't interchange with
> anything non-systemd.  Though as we can see from eudev/elogind and so
> on that isn't strictly the case.
>
> I sometimes describe systemd as the anti-busybox.
>
> But, I don't want to derail the thread entirely...
>
Poettering is very much "do it right, if something else is already
broken then just break it completely". Much better for a computer, but
ruffles peoples' feathers ...

That said, yes we are derailing this thread somewhat :-)

Cheers,
Wol

Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

Peter Humphrey-3
In reply to this post by antlists
On Monday, 11 November 2019 10:38:50 GMT Wols Lists wrote:

> Fact is, there are a lot of people out there who hate systemd because
> it's been successful, and it's been successful because it sticks to the
> nix philosophy of "do one thing, and do it well".

...but not the other *nix tradition of minimising keystrokes.

--
Regards,
Peter.




Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

Neil Bothwick
In reply to this post by Rich Freeman
On Mon, 11 Nov 2019 06:54:31 -0500, Rich Freeman wrote:

> > > It was obviously a semi-trollish comment.
> > >  
> > Now that's harsh! Although yes I'm sure he was tweaking tails - hence
> > the "tongue in cheek" smiley.  
>
> I didn't intend to suggest that I thought it was mean-spirited.  Just
> stirring the pot.

Guilty as charged ;-)

> > Fact is, there are a lot of people out there who hate systemd because
> > it's been successful, and it's been successful because it sticks to
> > the nix philosophy of "do one thing, and do it well".
> >  
>
> Now, THAT is a semi-trollish comment if I ever saw one.  :)
>
> That said, you could argue that the individual components of systemd
> do generally do one thing well.  I think the criticism is more in the
> packaging, and that the components mostly don't interchange with
> anything non-systemd.  Though as we can see from eudev/elogind and so
> on that isn't strictly the case.
It seems that most of the criticism is about the way it was developed and
by whom, rather than the merits or demerits of the code itself,

> I sometimes describe systemd as the anti-busybox.
>
> But, I don't want to derail the thread entirely...

I think you're too late :)


--
Neil Bothwick

Make like a tree and leave.

attachment0 (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: visualise openrc initscript start order and dependency tree

Neil Bothwick
In reply to this post by Peter Humphrey-3
On Mon, 11 Nov 2019 14:32:36 +0000, Peter Humphrey wrote:

> > Fact is, there are a lot of people out there who hate systemd because
> > it's been successful, and it's been successful because it sticks to
> > the nix philosophy of "do one thing, and do it well".  
>
> ...but not the other *nix tradition of minimising keystrokes.

Thank $DEITY for aliases!


--
Neil Bothwick

Remember, it takes 47 muscles to frown
And only 4 to pull the trigger of a sniper rifle....

attachment0 (849 bytes) Download Attachment