[gentoo-dev] Starting/stopping services

[gentoo-dev] Starting/stopping services

[Andy Arbon]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

I'd like to mention a problem that I see occasionally and see if anyone
else considers it annoying enough to have fixed somehow.

The problem is that services can't be stopped (via /etc/init.d/* ) if
the process they started has died since the init script started it.

Presumably the scripts try to terminate the process, find that it's not
there anymore and exit with an error, but this means that there is no
way (aside from starting a process with the correct name then letting
the script kill it) to start the service again until the system
restarts, which proves very annoying if you're doing anything which
causes services to die frequently (in my case I'm playing with the
lircd, which is a little flaky sometimes).

Is there anything that can be done about this? If the script tries to
stop a service and finds the process has died then it probably should
report an error, but I think it should also register that the service
has stopped so that it can be restarted.

Cheers,

Andy



-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQE+KZRqX3TTUvZURBERAu1/AKCOJ34Hz/aeL/piAOd2lOgPJSnnhgCfcogr
0D28blwIFmGEHKU+SfRs6O4=
=5djP
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Starting/stopping services

[Caleb Shay]

/etc/init.d/servicename zap

On Sat, 2003-01-18 at 12:52, Andy Arbon wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hello,
> 
> I'd like to mention a problem that I see occasionally and see if anyone
> else considers it annoying enough to have fixed somehow.
> 
> The problem is that services can't be stopped (via /etc/init.d/* ) if
> the process they started has died since the init script started it.
> 
> Presumably the scripts try to terminate the process, find that it's not
> there anymore and exit with an error, but this means that there is no
> way (aside from starting a process with the correct name then letting
> the script kill it) to start the service again until the system
> restarts, which proves very annoying if you're doing anything which
> causes services to die frequently (in my case I'm playing with the
> lircd, which is a little flaky sometimes).
> 
> Is there anything that can be done about this? If the script tries to
> stop a service and finds the process has died then it probably should
> report an error, but I think it should also register that the service
> has stopped so that it can be restarted.
> 
> Cheers,
> 
> Andy
> 
> 
> 
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.1 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> 
> iD8DBQE+KZRqX3TTUvZURBERAu1/AKCOJ34Hz/aeL/piAOd2lOgPJSnnhgCfcogr
> 0D28blwIFmGEHKU+SfRs6O4=
> =5djP
> -----END PGP SIGNATURE-----
> 
> 
> --
> gentoo-dev@gentoo.org mailing list
-- 
Caleb Shay <caleb@webninja.com>


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Starting/stopping services

[Burton Samograd]

[-- Attachment #1: Type: text/plain, Size: 809 bytes --]

On Sat, Jan 18, 2003 at 05:52:43PM +0000, Andy Arbon wrote:
> Presumably the scripts try to terminate the process, find that it's not
> there anymore and exit with an error, but this means that there is no
> way (aside from starting a process with the correct name then letting
> the script kill it) to start the service again until the system
> restarts, which proves very annoying if you're doing anything which
> causes services to die frequently (in my case I'm playing with the
> lircd, which is a little flaky sometimes).

If I understand what you are saying, this can be solved by giving the
init script the zap command (ie. /etc/init.d/lircd zap), which resets
its run state and lets you restart it normally.

-- 
burton samograd
kruhft@kruhft.dyndns.org
http://kruhftwerk.dyndns.org

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Starting/stopping services

[will]

They can't be stopped because there's nothing to be stopped.  However, zap
should work.

If you do "/sbin/runscript", then it'll tell you all about runscript 
stuff.

/will


On Sat, 18 Jan 2003, Andy Arbon wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hello,
> 
> I'd like to mention a problem that I see occasionally and see if anyone
> else considers it annoying enough to have fixed somehow.
> 
> The problem is that services can't be stopped (via /etc/init.d/* ) if
> the process they started has died since the init script started it.
> 
> Presumably the scripts try to terminate the process, find that it's not
> there anymore and exit with an error, but this means that there is no
> way (aside from starting a process with the correct name then letting
> the script kill it) to start the service again until the system
> restarts, which proves very annoying if you're doing anything which
> causes services to die frequently (in my case I'm playing with the
> lircd, which is a little flaky sometimes).
> 
> Is there anything that can be done about this? If the script tries to
> stop a service and finds the process has died then it probably should
> report an error, but I think it should also register that the service
> has stopped so that it can be restarted.
> 
> Cheers,
> 
> Andy
> 
> 
> 
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.1 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> 
> iD8DBQE+KZRqX3TTUvZURBERAu1/AKCOJ34Hz/aeL/piAOd2lOgPJSnnhgCfcogr
> 0D28blwIFmGEHKU+SfRs6O4=
> =5djP
> -----END PGP SIGNATURE-----
> 
> 
> --
> gentoo-dev@gentoo.org mailing list
> 

-- 
whatever it is, you can find it at http://www.bluesock.org/~willg/
except Will--you can only see him in real life.



--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Starting/stopping services

[Matt Tucker]

-- will <willg@bluesock.org> spake thusly:

> They can't be stopped because there's nothing to be stopped.
> However, zap should work.
> 
> If you do "/sbin/runscript", then it'll tell you all about runscript 
> stuff.

Is there any particularly good reason that stop doesn't just do this
automatically?


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Starting/stopping services

[Mike Frysinger]

this allows for better control/checking via scripts and what not ...

if theres an error/issue stopping a service, i want to know about it ... but 
if you dont care, just zap it ...
-mike

On Tuesday 21 January 2003 03:56, Matt Tucker wrote:
> -- will <willg@bluesock.org> spake thusly:
> > They can't be stopped because there's nothing to be stopped.
> > However, zap should work.
> >
> > If you do "/sbin/runscript", then it'll tell you all about runscript
> > stuff.
>
> Is there any particularly good reason that stop doesn't just do this
> automatically?
>
>
> --
> gentoo-dev@gentoo.org mailing list

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Starting/stopping services

[Dylan Carlson]

On Tuesday 21 January 2003 04:09 am, Mike Frysinger wrote:
> this allows for better control/checking via scripts and what not ...
>
> if theres an error/issue stopping a service, i want to know about it ...
> but if you dont care, just zap it ...
> -mike
>
> On Tuesday 21 January 2003 03:56, Matt Tucker wrote:
> > -- will <willg@bluesock.org> spake thusly:
> > > They can't be stopped because there's nothing to be stopped.
> > > However, zap should work.
> > >
> > > If you do "/sbin/runscript", then it'll tell you all about runscript
> > > stuff.
> >
> > Is there any particularly good reason that stop doesn't just do this
> > automatically?

I would like to have the ability to zap a service as "started".  There are 
some scenarios where a startup script will fire off many background 
processes and might fail on one or two of them.  

The script will be flagged as "stopped", but the background processes that 
are running will keep running.  So if there was way to set the script as 
"started", you could do a normal stop to kill off whatever did manage to 
start, instead of manually killing off processes.

Cheers,
Dylan Carlson [absinthe@pobox.com]

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Starting/stopping services

[J Robert Ray]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dylan Carlson wrote:
|
| I would like to have the ability to zap a service as "started".  There
are
| some scenarios where a startup script will fire off many background
| processes and might fail on one or two of them.
|
| The script will be flagged as "stopped", but the background processes
that
| are running will keep running.  So if there was way to set the script as
| "started", you could do a normal stop to kill off whatever did manage to
| start, instead of manually killing off processes.

This sounds like something that would be better solved by breaking the
script up into multiple init scripts for each background process, and
then set the dependencies as needed.

Are you referring to a package in portage or a custom job?

- - Robert

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.1.90-nr1 (Windows XP)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQE+LapEbv6Y11NqSv8RAqRIAJwKGbPDTi9pY+9i72pLrudw+0kXsgCfTdac
HNAY9LqonOtLrHBKWjg9u2k=
=F8Ca
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Starting/stopping services

[Dylan Carlson]

On Tuesday 21 January 2003 03:15 pm, J Robert Ray wrote:

> This sounds like something that would be better solved by breaking the
> script up into multiple init scripts for each background process, and
> then set the dependencies as needed.
>
> Are you referring to a package in portage or a custom job?
>

Either case...  One example, perhaps a stinky example (but I'll use it 
anyway) is the vmware startup script.  

IMHO, breaking some of these things up into multiple scripts is 
overcomplicating it.  Especially if things need to fire in a certain 
order.   It would be easier for me to just have one script  to zap as 
either started or stopped.... zapping started would allow me to simply 
shut down any parts that happen to be running without hunting around for 
PID's.  

Although I can see where splitting it into several separate init scripts 
would give more granular control, sure.

Cheers,
Dylan Carlson [absinthe@pobox.com]

--
gentoo-dev@gentoo.org mailing list