[talk] In Contrast to BSD init, A history of modern init systems (1992-2015)

Pete Wright pete at nomadlogic.org
Mon Sep 21 18:07:11 EDT 2015

On 09/21/15 15:02, Edward Capriolo wrote:
> " developer has written a daemonized processes.  it's buggy and crashes
> periodically, but we don't have resources at the moment to fix the bug
> causing the daemon to crash because corporations are bad and should feel
> bad.  we used daemontools to monitor that the deamon was running, and
> had it automatically restart if it crashed."
> ' We all have done this, but this is a very dangerous path. I worked at
> a place where I was a sysadmin. I was BLAMED because my 'restart
> processes' failed to properly restart 'buggy developer process x'. I
> always consider this the first step on the highway to hell (aka Janitor
> ops). Once the enterprise adopts the notion that 'smart developer types'
> are 'too busy' to make processes that do not crash or do not fill disk
> with logs, all the crap work falls on ops. Soon people sending you tar
> files because they 'too busy to make installer'. 
> Flashbacks sorry.
> In the end having something restart downed processes is smart but if
> anything starts crashing repeatedly something need to be done. IE 1
> random crash a week means someone go fix it!

oh yea i %100 agree with you about how horrible this use-case is.  as a
sysadmin i'm sure you are also familiar with inheriting
huge/complex/horribly-written systems which was held together by
duct-tape, daemontools and nightmare inducing /bin/sh scripts.  this was
one of those guys.

and after going through that rodeo at several different jobs i came up
with my opinion that technical debt should *not* be hidden, but put
front and center so that it gets addressed :)


Pete Wright
pete at nomadlogic.org
twitter => @nomadlogicLA

More information about the talk mailing list