Forum OpenACS Q&A: Anybody using inittab? Should it stay in the docs?

Collapse
1: Anybody using inittab? Should it stay in the docs?
Posted by Joel Aufrecht on 03/09/03 08:03 PM

I propose removing inittab documentation from the install docs, leaving only the daemontools documentation for service automation.

Arguments for removing it

  • It involves mucking with inittab, which is risky
  • It has very few features. It keeps the server up, period. You can't take the server down without again mucking with inittab.
  • It doesn't log
  • It's platform-specific

Arguments for leaving it in

  • Daemontools requires additional steps and software install to set up
  • Some people may just prefer inittab, or are more familiar with it from previous installs, and want current documentation.
Collapse
2: Re: Anybody using inittab? Should it stay in the docs? (response to 1)
Posted by Don Baccus on 03/09/03 08:34 PM
I think we should document both approaches.  Plenty of old-timers like me still use inittab and documenting it doesn't really cost us anything, does it?
Collapse
3: Re: Anybody using inittab? Should it stay in the docs? (response to 1)
Posted by Tom Jackson on 03/09/03 09:12 PM

As long as the method works I think we should keep the documentation.

Collapse
4: Re: Anybody using inittab? Should it stay in the docs? (response to 1)
Posted by Andrei Popov on 03/10/03 08:17 AM
FWIW, daemontools have been causing me nothing but trouble on FreeBSD -- it would spawn multiple processes until things will grind to a [not really a halt] VERY slow motion.

While I understand that FreeBSD is not likely to be one of "reference architectures" it is definitely a "supported system", hence I'd vote for leaving inittab section.  If one wishes, daemontools can be specified as a suggested approach, though.

Collapse
5: Re: Anybody using inittab? Should it stay in the docs? (response to 4)
Posted by Joel Aufrecht on 03/10/03 08:40 AM
I had the same trouble with daemontools at first, including with the runaway processes. Like all djb software, it's very unforgiving of human operators. OTOH, it is very well suited to multiple OpenACS services on one machine, and it's easy to start and stop if it's installed exactly correctly.

The current draft walks the user through a daemontools configuration and includes a link to the inittab method, which is now documented outside of the main flow. If this isn't acceptable, please let me know.

Collapse
6: Re: Anybody using inittab? Should it stay in the docs? (response to 1)
Posted by Lamar Owen on 03/12/03 12:17 AM
I have used both inittab and straight initscripts.  There are of course dangers to initscripts; but many things are already set up that way.

Init is _Designed_ to do this stuff, and, if you're careful (or have console access to the machine in question) there's not danger in munging inittab.