[Nut-upsdev] DNS-SD (was Re: [nut-commits] svn commit r2974 - branches/nut-scanner/tools/nut-scanner)

Arnaud Quette aquette.dev at gmail.com
Tue Apr 19 14:26:04 UTC 2011


Hi Charles,

2011/4/19 Charles Lepple <clepple at gmail.com>

> On Apr 19, 2011, at 7:44 AM, Arnaud Quette wrote:
>
>  Author: aquette
>> Date: Tue Apr 19 11:43:46 2011
>> New Revision: 2974
>> URL: http://trac.networkupstools.org/projects/nut/changeset/2974
>>
>>  ...
>
>> +void scan_nut_avahi()
>> +{
>> +       printf("Scanning NUT bus (DNS-SD method):\n");
>> +
>> +       /* Check avahi-browse code:
>> +        *
>> http://git.0pointer.de/?p=avahi.git;a=tree;f=avahi-utils;h=5655a104964258e7be32ada78794f73beb84e0dd;hb=HEAD
>> +        *
>> +        * Example service publication (counterpart of the above):
>> +        * $ avahi-publish -s nut _upsd._tcp 3493 txtvers=1
>> protovers=1.0.0 type=standalone
>> +        */
>> +}
>>
>
> Just some thoughts:
>
> It's been a while since I last looked at the DNS-SD spec, but we might want
> to use either "_nut._tcp" (following the IANA reserved port name) or
> "_nut_upsd._tcp" because there is at least one other "upsd" in Debian alone.
>

indeed. this is part of the things that need more reflexion.
for the time being, it's still an early proof-of-concept, but the "nut Vs
upsd" notion is already in the pipe.
what still puzzles me is that "_upsd._tcp" is used as Service Type.
while "nut" is used as the Service Name.
(instant update) I've just made a quick test with Frédéric, and the "nut"
name *must* be unique: each system that publish a nut service availability
has to have a unique Service Name!

this is something I'll be discussing with Ubuntu folks at the next UDS
(since they are (ab)using avahi for cloud deployment).
in the meantime, feedback / comments / ideas are very welcome.

Also, there are XML files that we can create in the Avahi configuration
> directory to publish that information. This might be easier than adding
> avahi-publish to all of the various startup script types that distributions
> have these days.
>

indeed. The show stopper here is that I want to publish the service
availability *only* if MODE is different that "none" (so if nut is actually
running).
and I'm not yet sure if we can put condition on the publication through xml
(I don't think so), while the initscript modification would be limited to
executing the above "avahi-publish" line with "type=$MODE"

Note that I'm still unsure on this too, since publishing unconfigured NUT
may be a use case in infrastructure: Augeas and Puppet will allow remote
configuration, which is in favor of the XML publication.

If you don't want to pull all of the Avahi library code into the scanner,
> many systems have the "avahi-browse" command-line tool.
>

yup, this is what I used during the very first stage:
$ avahi-browse _upsd._tcp

As you can see, this is still a WIP...

thanks a lot for your comments Charles, and don't hesitate to bounce on the
above...

cheers,
Arnaud
-- 
Linux / Unix Expert R&D - Eaton - http://powerquality.eaton.com
Network UPS Tools (NUT) Project Leader - http://www.networkupstools.org/
Debian Developer - http://www.debian.org
Free Software Developer - http://arnaud.quette.free.fr/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/nut-upsdev/attachments/20110419/566b4c4b/attachment.htm>


More information about the Nut-upsdev mailing list