[CentOS] advice on using daemontools on centos production servers

Mon Oct 25 10:54:39 UTC 2010
saurabh verma <whatnotsaurabh at gmail.com>

Hi advisors , 

I would some suggestions on using http://cr.yp.to/daemontools.html .

My requirements is to use less systems daemon checks in nagios but concentrate more on  business / application level checks . 

We have lots of daemons like below 

> 1. ) snmpd
> 2.) ntp
> 3.) SSH
> 4.) crond
> 5.) php-cgi
> 6.) nginx
> 7.) Postfix
> 8.) wackamole
> 9.) spread
> 10.) atop
> 11.) redis
> 12.) taillog-dopa
> 13.) taillog-user
> 14.) rsyncd

... ETC. 

We use nagios checks + event-handlers in case of daemon failure . 

My views is to use daemontool to control all these daemons , and create a single nagios check which scans /service/ directory and alerts for daemons which are getting restarting in unwanted ways. 

Some of my friends will argue about daemontools being non-unixy and unconventional , But I had good experience with daemontools on around 3000 servers earlier in my previous organization . 

Please advice what you guys think about daemontools and its pros and crons . 

Thanks , 
Saurabh Verma