DHCP - fail only difference is Netonix switch

[anthonyw]

This is an odd one, wondering if anyone else has seen this.
We have a DHCP server that will fail to provide an updated IP. This has been an on going issue on one segment of our network and it happens so rarely we just deal with it and move on. But the exact same issue is now appearing on a completely different and isolated section of our network. The only core / tower equipment change on this segment is swapping a toughswith pro with a netonix DC-8 93 days ago.

[lligetfa]

If I had to guess (and I must), I'd say probably a flat network and broadcast rate limiting is throwing away some DHCP packets.

[anthonyw]

Good guess but where is this limit at? I don't have it set in any AP's or the netonix. It is a flat network with about 500 or less devices per segment. I keep thinking it is a bridge table getting filled as it appears to effect ubnt equipment prior to any other hardware. ubnt is not known for it's quality so...
But I have many different platforms on these segments and like in the original post the only change other then adding user devices CPE is the netonix.

[sirhc]

Well so many times coincidence leads people astray.

Simply put the TS PRO back in and see if the problem goes away.

If problem goes away then post up screen shots of all the TSPRO config and the WS config and we will go from there.

If the problem does not go away then you know something else if a foot.

You should be running latest version v1.5.1rc12 as that is the current latest code and what is being developed.

Also a large FLAT Layer 2 segment with 500 devices is not a good idea.

[anthonyw]

Agreed, coincidence is a pain, why I am asking if anyone else has seen the issue. I have been fighting this over 18 months. It only crops up every month or two for a day or two and then goes away. Driving me crazy that it has spread to a new segment and the only change was the netonix. So the reason for the question. Not asking for "support" per say. My next step is to remove the netonix from the original issue segment and see if the issue resolves.

[lligetfa]

Good guess but where is this limit at? I don't have it set in any AP's or the netonix...

Limit would be set under storm control but then you say you did not set any limit. Another guess is that discovery is enabled on the switch and contributing to an already problematic broadcast issue. If not that, then my next guess is pause control. After that, I'm out of guesses.

[anthonyw]

I'm sorry, attached is the setting tab so you can stop guessing :) .
All of that is shut off as is flow control, MS and DS on the ports tab.

[sirhc]

Agreed, coincidence is a pain, why I am asking if anyone else has seen the issue. I have been fighting this over 18 months. It only crops up every month or two for a day or two and then goes away. Driving me crazy that it has spread to a new segment and the only change was the netonix. So the reason for the question. Not asking for "support" per say. My next step is to remove the netonix from the original issue segment and see if the issue resolves.

Since it happens infrequently my guess is it has nothing to do with the Netonix which should be confirmed when you put TS back in for awhile.

This is the problem with LARGE Layer 2 segments, very hard to find something when it goes bad.

I used to run large layer 2 flat network, best thing I ever did was put routers are all towers and segment each AP to its own segment with subnetted IPs using the switch and VLANs to provide the needed ports.

Read my post here: viewtopic.php?f=30&t=452#p2961

Watch the movie too

[anthonyw]

Thats funny since you sell a device designed to make "large" flat networks.... maybe you should stop selling something you don't believe in? As far as I can tell, nearly all your improvements to the line are for making even larger flat networks.
And with that said, what is a "large" flat network in your most humble opinion?

[sirhc]

Thats funny since you sell a device designed to make "large" flat networks.... maybe you should stop selling something you don't believe in? As far as I can tell, nearly all your improvements to the line are for making even larger flat networks.
And with that said, what is a "large" flat network in your most humble opinion?

No, I do not endorse and never have endorsed large Layer 2 Networks for WISPs it is a management nightmare and when something goes wrong like a faculty device, loops, broadcast storms, and many others tracking this down becomes a nightmare.

A POE switch is a tool, how you use the tool is up to you. If you went and read my post I linked above I show how my WISP network is setup and how I use Netonix. There are many ways to do a networks and large flat layer 2 segments is one and works great until it doesn't then trying to find the culprit is hard.

Most larger old WISPs will tell you they have or are moving away from LARGE flat network segments towards smaller layer 2 segments with routing between the segments because when something goes wrong in a LARGE layer 2 segment finding the problem is very difficult and effect a lot of customers.

With smaller layer 2 segments problems affect smaller # of devices and customers and pin pointing the issue is a lot easier and faster.