Bug 30113

Summary:	bind initscript 'reload' function does not work
Product:	[Retired] Red Hat Linux	Reporter:	Elliot Lee <sopwith>
Component:	bind	Assignee:	Bernhard Rosenkraenzer <bero>
Status:	CLOSED RAWHIDE	QA Contact:	David Lawrence <dkl>
Severity:	medium	Docs Contact:
Priority:	high
Version:	7.1	CC:	harald, pbrown
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2001-03-01 19:41:57 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Elliot Lee 2001-02-28 23:46:23 UTC

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux 2.2.16-24smp alpha; en-US; 0.8)
Gecko/20010217


the 'ndc' program doesn't work because named isn't listening on the UNIX
socket.

Reproducible: Always
Steps to Reproduce:
1. service named reload

Can also run "/usr/sbin/ndc reload" to see the actual error message.

Actual Results:  error message

Expected Results:  named restarted

Proper solution is probably 'kill -HUP <named pid>', so that listening on
the UNIX socket isn't required.

High priority because bindconf depends on this behaviour

Comment 1 Bernhard Rosenkraenzer 2001-03-01 10:11:59 UTC

The "ndc" program is obsolete and has not been included with bind starting 
with version 9.0.0, which has been in the tree for months.

the init script uses the correct command, /usr/sbin/rndc reload, which works 
with the current bind package.

If it doesn't work, it's a configuration problem (you probably didn't set up 
rndc keys).

Comment 2 Elliot Lee 2001-03-01 18:24:37 UTC

rndc does not work either ("connection refused").

It is unacceptable to require the user to perform bind configuration in order
for the user to perform bind configuration! :) The initscripts must function
properly without setting up servers, keys, or any other fancy stuff. 'service
named reload' must work out of the box. This is the way of the Jedi...

I am not Glen, but I personally consider this a MUST-FIX for 7.1, since a
package that I have worked on (bindconf) requires this to work correctly in
order for configuration changes to take effect.

Comment 3 Glen Foster 2001-03-01 18:31:45 UTC

Preston, I concur with Elliott's assessment -- do you agree this is MUST-FIX?

Comment 4 Bernhard Rosenkraenzer 2001-03-01 19:41:53 UTC

I'd still say this is NOTABUG: The concept of running *reload* on a service that 
hasn't been configured (and therefore shouldn't be running) is simply asking for 
trouble and security problems.

service anything reload

should never work on a service that isn't running already, and an unconfigured 
service shouldn't be running.

I can switch over to the kill -HUP hack - no problem, but I think this introduces a 
bug rather than fixing one.

Comment 5 Bernhard Rosenkraenzer 2001-03-02 08:32:03 UTC

It's changed in 9.1.0-6, but I still think reloading an unconfigured service
shouldn't work.

Comment 6 Wouter Liefting 2001-06-13 21:35:56 UTC

The problem is partly a configuration error too. Try adding the following to /etc/named.conf:

controls {
  inet 127.0.0.1 allow { localhost; };
  keys{ "key"; };
};

This actually allows rndc to connect to the named daemon properly.