Re: starting service order

[Sebastien ESTIENNE]

Marco Ermini wrote:

> <quota chi="Sebastien ESTIENNE">
>  
>
> > > To provide such a service, you should define:
> > >
> > > 1) the ability for a dependency to fail
> > >
> > >
> > >      
> > Yes that's i was asking if this feature exist, it could be implemented
> > like this
> >
> > depends on apache_bin can fail
> > or a new keyword that would do the same thing named "before"
> >    
> </quote>
>
> This would sound cool.
>
>
> <quota chi="Sebastien ESTIENNE">
>  
>
> > > 2) a kind of ordering, or at least, the ability to not startup the
> > > subsequent service in the list if the previous did not already completed
> > > to startup _without_ using a dependency
> > >
> > >
> > >      
> > it could be implemented by keywords like before/after
> > Gentoo linux already provide this functionnality (look here:
> > http://www.gentoo.org/doc/en/handbook/handbook-x86.xml?part=2&chap=5#doc_chap4
> > )
> > you can define order with before after
> > dependencies need/use and also provide
> >
> > for example you can say that qmail "provide mta" and "need  net"
> > and that nagios "need mta"
> >
> > so the system knows that he must start qmail before nagios (if you had
> > use the "use" keyword instead of "need" it means that qmail can fail
> > it's okà
> >
> > "provide" is also because you can say that many service provide the same
> > functionnality, eg: mta -> qmail, postfix, exim
> >
> >    
> </quota>
>
> No, I never used gentoo (that's a fault, actually, since I think I sould
> try it...). Just having a quick look at that web page (so don't think I am
> actually "judging" or something similar, it's just the first impression of
> a Unix sysadmin - and I _can_ change my mind, I'm not that kind of
> sensitive :-), it sounds like it's a way to manage system startup too much
> complicated for an Unix sysadmin' "average" taste; it seems like it adds
> complexity that it's not worth the cause. When I configure an MTA I want
>  
in fact it doesn't add complexity but choice, because on unix many 
software provide the same functionnality:
eg: syslog, metalog, syslog-ng -> logger
fcron dcron vcron -> crontab
postfix , exim, qmail -> mta

so this dependecie system can work with any "brick" as long as it 
provide the same functionnality.
In my opinion it's great for sysadmin, because every sysadmin has his 
fav tools, and in gentoo, doesn't assume any default tool, so that every 
tool are well integrated in the system, for example in redhat postfix is 
the default, and it works great in redhat but if you want to put qmail 
instead, you have quite some work to make it integrates as well as 
postfix, in my opinion that's why gentoo is a meta-distribution, and 
that it can fits different need, because they assume no default, that's 
the same for the logger, the cron, the db, and many other things i 
forget about, ah yes, Java, in gentoo, you can use whatever jdk you want 
(sun, ibm, blackdown, ....) you can even install the 3 and then choose 
which one you want to use with a tool (java-config, the system jvm can 
be different than users 'jvms).
And i'd like to integrate monit in the same spirit, i don't want it to 
replace the current system, i want it to be an alternative, and (if 
possible) a complement of the default system, to assist it in monitoring 
system.

> it to work and work the way I want: I don't want a "replacement" MTA to
> start if my MUA fails, since I spent time configuring _my_ MTA, the MTA I
> _choosed_, with my blacklists, anti-spam etc. etc. - and it's actually a
> _cost_ to configure two different software in the same way "just in the
> case"...
>  
sorry i think i wasn't clear, it was not about configure 2 mta, it's 
about to be able to choose the one that fits your need, and keep the 
system still fully functionnal whatever mta your choosed

> Finally, I think that your idea would add some level of complexity to
> monit, which may be useful, but I don't know if the authors would like -
> let's see what they think about that :-)
>
>  
yeah it would be great to have their opinion about this :)

> <quota chi="Sebastien ESTIENNE">
>  
>
> > > 3) some tool to easily rearrange the services (a chkconfig-like tool) 4)
> > > the ability tu support different runlevels
> > >
> > >
> > >      
> > I wrote this tool, it was easy, i just modified the default tool from
> > gentoo (it was just a matter of 5 lines)
> > I did it like this:
> > in /etc/monitrc i added  "include /etc/monit.d/autostart/*"
> > in /etc/monit.d/ i put a file by service (eg: apache postfix mysql)
> > and the tool just write or remove a link from /etc/monit.d/ in
> > /etc/monit.d/autostart
> > so i have the exact same functionnality as a tool like chkconfig
> > (i don't know if you use gentoo, but now i add services like this:
> > rc-update add apache monit)
> >    
> </quota>
>
> I didn't know about how gentoo works, but this sounds interesting: adding
> and removing a service from monit with a command it could be useful... but
> maybe you can just unmonitor it? :-)
>  
i thought about this, but it was so much easier to do it this way, and 
it is also much more consistent why how gentoo does things.
Gentoo is different from redhat/debian in the the way they don't care 
about sysV runlevel 1-6, runlevels have a name and you can define as 
much runlevel as you want, (even there is 2/3 default one)
the standar one is "default" so i just added one called "monit", that 
way you choose if you service is manage with the "default" system or the 
"monit" one :)

> Anyway it seems to me this is just a way of using monit: adding or
> removing a file from a directory is just a matter of a 5 line script (like
> you already said) and this did not require any modification to monit - I
> think.
>
>  
i don't modification in monit for thiss functionnality, the "includes" 
work perfectly, i'm just missing "order" functionnality, or just being 
able to say that a dependencie can fail. nothing more

> <quota chi="Sebastien ESTIENNE">
>  
>
> > you said that monit is only good in specialized situation, or when
> > customized, it's true, but you can define default and then allow the
> > user to provide custom settings without modifying the default install,
> > using include:
> > the default, would only check for pid, and listening port and then the
> > user can overide things or adding things by including his configuration:
> > eg:
> > /etc/monit.d/system/SERVICE <- contain all default service config
> > and each SERVICE contain a line like this at the end: include
> > /etc/monit.d/user/SERVICE
> > so when a user want to customized apache for example, he just have to
> > create a file /etc/monit.d/user/apache
> >
> > I think that monit can remplace the default system, because the onlyt
> > things that the default init system has to do, is stating/stopping
> > system in the right order nothing more
> > and i don't see why monit couldn't do this?
> > for example djb's daemontools where built to manage services (they
> > manage qmail/djbdns) and monit seems to be a daemontools++. too bad he
> > also missed the ability to define services start order...
> >    
> </quota>
>
> The first thing I usually think is that a program should serve the purpose
> it was being created for. The "Unix way" of thinking is: simple tools to
> be put together in a simple way. Changing Monit to being more complex to
> serve a different purpose for what it was conceived for, sounds to a
> "Unix" mind like mine (a little simple and orthodox, I think ;-) like some
> kind of "noise" I don't need to listen to ;-)
>
>  
in fact it's my mistake i thought that monit was designed to a be 
possible init replacement

> Anyway you can do it... adding a feature to monit it's not forbidden by
> the law ;-) and two of the features you suggested (dependencies which can
> fails, and the order of startup) could be useful to the "conventional" way
> of using Monit, too (I think).
>
> I think that there is a much more useful option which is missing: the
> "startup in progress" option... when a service is long to startup (think
> about Java application servers, ecc.), Monit becomes unresponsive if you
> ask for a "status".
>
>  
thanx for pointing this, now i understantd why "monit status" sometimes 
fail :D

> I think that starting a service should spawn a different thread and
> changing the status in "startup in progress" (this would need to be
> configurable from the monitrc file). This could also be useful in a
> "system startup" situation like you want for gentoo, as it would speed up
> the startup of the whole system if you have services which didn't need to
> be "dependable" from each others.
>
>  
In fact gentoo already speed things up, because it can start services in 
parallel when they don't depend on another one.

but you are right "starting in progress feature" and threading would be 
really usefull features too!.