Improving the host <parents> logic

Shane Stixrud shane at geeklords.org
Fri Dec 16 01:55:27 CET 2005


On Fri, 16 Dec 2005, Andreas Ericsson wrote:

>> Are you serious?  You consider this the Right Thing (TM) compared to having 
>> the ability of defining the layer2 and layer 3 parent for each device?
>
> Yes. Most problems I've seen with VLANs doesn't so much have to do with the 
> management network or physical device being down, but rather that one or more 
> interface either breaks or go out of tune for one reason or another. Checking 
> the management interface in such cases only catch a few of the possible 
> errors.
>

We are not talking about whether pinging the management interface will 
catch every "switch problem" (although I would say it does in 50% of the 
cases).

You said and I quote "Give the switch an IP in each network and make it 
switch2-vlan23", suggesting this as a solution to the parent logic 
limitations, here you are discussing check_command logic which is totally 
unrelated.  That being said if the layer2 and layer3 parent relationships 
were in place it is simply a matter of defining services that check for 
more important criteria than just pinging the management IP.  For example 
checking to see if the switch port in question is inactive, failed or if 
the switches VLAN exists/is in a fault state.

> Since you asked: I've used HP, Nortel, Extreme, D-Link, Cisco, Netgear, 
> Zyxel, TekComm (don't use those), 3Com and Linksys.
>
> All of those, with the exception of TekComm, supported multiple IP-addresses. 
> All of them also had the option of only serving tcp and udp-based services on 
> one IP only.

Interesting, as far as I know up until recently Cisco switches were 
limited to sc0.  Also I note that you have changed tunes from saying its 
possible to assign an IP addresses on every switch for each vlan to 
saying they support "multiple" IP address, which some do and some don't. 
In either case such an approach is silly at best.

> I have no influence what so ever what code gets in and what doesn't. I 
> doubt if Ethan will write the code for this  though, so if you want it 
> in you should start hacking right about now.

Did not mean to imply you did.  I appreciate your opinion on what Ethan 
will find interesting and worth while.  I wish I were more proficient at 
C but I am not, my programming skills are limited to relatively 
simple python programs for the most part. It may be possible for funding 
to be obtained for a feature like this however.

>> Every inter-network device has at least one layer 2 connection and at least 
>> one layer3 interface.
>> 
> A straight answer, if you please; Do you intend for host-objects to support 
> the three configuration directives "parents", "l2parents" and "l3parents"?

I see no reason why l3parents wouldn't function exactly as parents 
functions today, at least from a user perspective.  That being the case 
only two parent directives would need to exist.  For those who do not 
want to define l2parents they would only need to define the l3parent 
directive maintaining existing functionality. Also I see no reason why the 
"parents" directive could not remain as an alias to l3parents.

Cheers,
Shane


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click




More information about the Developers mailing list