Reply

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

PDunnington
Luminary

M4300-12X12F replacing M7100-24X - Wireless Access Points not working

Hi Everyone,


I'm attempting to replace our ageing M7100-24X with a M4300-12X12F (XSM4324S) which I bought just over two weeks ago. I haven't contacted support yet as I'm hoping that perhaps someone has had a similar issue and a quick fix.


The issue I'm having is that our Extreme Wireless Access Points don't work, or rather are not detected by the Wireless Controller when plugged either directly or indirectly into the new M4300-12X12F.


Just as a bit of background; we use almost exclusively Netgear Switches with 4 VLANs for LAN access, General Wireless Access, Staff Wireless Access, Guest Wireless Access & VOIP. We use an Extreme VX9000 virtual Wireless Controller and various model Extreme Wireless Access Points.


The M7100-24X is our core network switch connected (VLAN trunked) to an adjacent GS728TS(B) and ultimately the plan is that the M4300-12X12F will replace both in the same role at the centre or our LAN in our company's Server Room.


After transposing all the relevant settings (VLANs etc.) from the M7100 to the M4300, and trunking a link between them, I attempted to migrate our LAN links to the new M4300 during a relatively quiet period.

Most services worked as anticipated; VOIP VLAN, Server access, Internet access, Cloud applications. But all our Wireless Access Points were no longer detected by our Wireless Controller. I immediately checked that I had got the VLAN configuration correct and confirmed that it was identical to the M7100. Flummoxed, I moved the LAN links back to the M7100 and the WAPs

all started operating correctly again.


I decided to test further by bringing a WAP from an outlying area and plugging it directly into the core switches to observe its behaviour first hand. Again, on directly connecting the WAP to either the M7100, GS728TS(B) or a GS110TP that I had available for testing; the WAP connected correctly and was successfully adopted by our Wireless Controller.


When directly connected to the M4300, again it failed to be detected or adopted, but the WAP did correctly receive a DHCP address from our DHCP Server and was "pingable" from the Wireless Controller and other LAN devices. The only logical conclusion I can make is that the M4300 (being a newer and ultimately more secure device) is either "blocking" or treating differently some protocol that is used to verify the presence of our WAPs.


As a final attempt I factory reset the M4300 and again connected the WAP with no VLANs set (our Wireless Access Points use the default VLAN1 as the management interface so should connect in this scenario, albeit without any Wireless functionality) - again the same result, no detection or adoption of the WAP.


I've also tried various Firmware versions; from the initial release 12.0.0.14, to the newest 12.0.17.6 and one somewhere in the middle. All with the same result.


I was initially very impressed with the M4300, but as it stands for us it's a bit of a "chocolate teapot".

Hoping someone can give me some guidance on the next steps to resolve this.


Thanks & Kind Regards,

Paul

Message 1 of 10

Accepted Solutions
schumaku
Guru

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

Paul,

 

As you mentioned the data is flowing over that MiNT tunnel between AP and controller - that is making use of 24577/UDP (wally brain is awake again) - I guess trouble with discovery and management is more on that special ether-type 0x8783 for L2 than on UDP used for L3 (routed connections).

 

Some sniffing would easily proof if the switch does fail to deal with ether-type 0x8783. That's why I point on the Netgear switch engineering resp. fellow Laurent Masia. 8-)

 

Regards,

-Kurt

View solution in original post

Message 9 of 10

All Replies
msi
Luminary Luminary
Luminary

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

Your post is quite long, but it doesn't contain information such as a example port config or a small diagram to more easily spot things you might have missed in transplanting from the M7100 by the M4300 . The M7100 is older, but it should (?) not have more feature than the M4300. It could be helpful to attach (not copy-paste) a sanitized config from the M7100 and M4300 (and also highlight which ports relevant / connected to what item).


I could imagine you are running into something you just didn't spot, it happens, but the text config could be helpful to make a comparison. 😉 If your Extreme APs have a serial console, you might also be able to see what they are missing if you can see L2 connectivity as you say. Maybe they spit out error messages that might give you a clue. I once hat issues with Cisco APs refusing to join theior WLC unless both APs and WLC had access to a proper NTP time source at startup.

Message 2 of 10
PDunnington
Luminary

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

@msi 

Hi MSI,

 

Thanks very much for your reply and apologies for my delay in replying; I've been away from work for 10 days.

 

While you're correct that it's very possible that I missed some configuration moving from the the M7100-24X to the M4300-12X12F, I didn't go into specific Port configurations as at that point I'd actually assumed I'd made a config error and Factory Defaulted the M4300 to again start from scratch.

This, I suppose is the crux of my issue; A Wireless Access Point (which uses the default VLAN 1 as it's management VLAN) directly connected to the Factory Defaulted "vanilla" M4300-12X12F is not adopted or "seen" by the Wireless Controller.

The same Access Point when directly connected to the Factory Defaulted M7100-24X (or any other of our numerous Netgear Smart Switches) is correctly adopted & "seen" by the Wireless Controller. The AP still needs configuration to work correctly in our environment, but it's presence and connectivity on our LAN is established. I reiterate; this is before any configuration changes are made from Factory Default.

This is quite a funadamental difference in behavior, one I haven't come across before, and one I'm struggling to understand.

 

I've attached a document showing a simplified diagram hopefully helping to clarify what I've said.

 

Thanks & Kind Regards,

Paul

 

Message 3 of 10
msi
Luminary Luminary
Luminary

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

Hi

OK, it's interesting that you are using a lower-spec GS724 as a core switch but a much beefier M7100 for your APs, nonetheless. There are a couple of variables to check, but without configurations provided, it's not much more than a lightly educated guess still.

 

You didn't mention the AP model so I can't quickly check if it would have a serial connection available, as mentioned, this could actually be helpful in order to identify potential issues.


Unless I'm completely wrong, this M7100 doesn't provide any PoE, the M4300-12X12F definitely doesn't provide any PoE at all so I can only imagine you are powering the AP from an external power supply. - To me that's another rather unusual decision since powering APs via an external power source adds more cabling. (However due to this we can exclude issues related to the PoE budget allocated to the AP since both switches do not provide PoE.)

It might also well be that your Extreme APs have some weird issues with autonegotiation, I doubt your APs support 10GBASE-T ports. Usually a non-working autonegotiation leads to no connectivity (you said you could ping the AP from the controller), so while unlikely you could force i.e. 1G on the switch port side.

 

If both the M7100 and the M4300 are reset to factory default, you could check the port configuration on the core switch GS724, or if possible, simply connect the M4300 into the port the M7100 is currently connected to. While most GS* switches only have a web UI, you can export a text config - you should be able find a per port section where you can usually more easily spot differences in the port config.

 

Use the "show logging buffered" when connected via SSH, Telnet or Serial on both the M7100 and M4300. Can you spot error messages when the APs gets plugged in? You can also check if the port status is really on the AP and uplink sinde on the M4300 using the "show interfaces status all" command. Then you can also check with "show mac-addr-table" if you can see the MACs of both the WAP and the controller.


But again: They are only educated guesses. Providing logs, Model types and configuration could help.

Message 4 of 10
PDunnington
Luminary

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

@msi 

Hi MSI,

 

Thanks again for your reply and suggestions, they're much appreciated.

 

Again, I should apologise for not being as clear as I intended or thought I was being:

The GS724TS(B) is only temporarily (hopefully) being used as the centre of our core LAN because the M7100-24X had began to behave erratically and the M4300-12X12F which was supposed to be it's replacement, won't work in the same way (This is where we came in).

 

Also I should explain that the diagram I provided was that of a "diagnostic" setup I used to try to illustrate that it was the M4300 that was behaving differently to our other Netgear Switches when placed in the same role/location. Our APs are not normally connected directly to these switches, they are normally located in factory/production areas which are located up to 6 "hops" away through other Netgear switches (mainly GS748Tv5 & GS724Tv4 & GS724TPv2 at the "end of the line" which provide POE for the APs).

You are correct that in this "diagnostic" setup I had to use a POE injector for the AP as none of the switches provide POE, although this didn't affect correct operation except through the M4300.

I should also say that it was the same AP connected to each of the switches in turn not 4 APs concurrently as my diagram may have suggested. Again, apologies for the lack of clarity, I was just trying to demostrate the consistency in behavior between all the switches (even the lowly GS110T) in contrast to the M4300.

 

Some of your suggestions I've tried; using a Syslog server on both the M7100 & M4300 when the same AP is connected; both switches' logs only show the port coming "up" but no errors, and both the AP's and Wireless Controller's mac addresses show up correctly in the address tables of both switches.

The AP model is a Extreme AP410e which indeed does have a console port so I'll attempt to view the APs startup via putty whilst connected to the different switches to see if I can discern any difference from the AP side.

 

Thanks & Kind regards,

Paul

 

 

 

Message 5 of 10
PDunnington
Luminary

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

@msi 

Hi MSI,

 

Some additonal information which I was able to discover this morning:

 

Although the WAP is still being reported by the Wireless controller as being offline, my Laptop is able to connect via a SSID broadcast by the WAP. The WAP is connected as before to Port 13 of the M4300 - XSM4324S.
My Laptop correctly receives a DHCP address and is able to ping both our gateway IP and the WAP IP which is also received correctly from our DHCP server.

 

Kind Regards,

Paul

Message 6 of 10
schumaku
Guru

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

The problem is that these Wireless WING controllers and AP are making use of a very strange architecture. Both disovery/management between the APs and the controllers are handled very different than "normal" network traffic the civilised network world is thinking of - for whatever crazy design ideas.

 

eg. FMI: https://extremeportal.force.com/ExtrArticleDetail?an=000086596

 

When I remember right, also the effective traffic does not flow over the SSID-associated VLAN, much more they use some fancy tunneling, so yet another UDP port in use (somewhere in the 25xxx range). As this traffic "tunnel" is just normal UDP, this might allow letting the traffic flow as you report, but not the discovery and management.

 

Needless to say, it's possible the M4300 does not let freely pass non-common frames. This is an issue Netgear switch engineering has to look into.  @LaurentMa

 

Message 7 of 10
PDunnington
Luminary

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

@schumaku

Thanks for you reply.

 

Yes, that makes sense and reading the document that you linked to confirms what you've said about the unusual way WING works with regards to AP discovery and control.

Based upon that (and I realise I'm putting you on the spot here); if you had to hazard a guess, would you say the M4300 is more likely to be "falling down" on passing the specific Layer 3 UDP port 24576 traffic or the Layer 2 ether-type 0x8783?

Or is that something only the Switch Engineering team would know? 

 

Many thanks,

Paul

Message 8 of 10
schumaku
Guru

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

Paul,

 

As you mentioned the data is flowing over that MiNT tunnel between AP and controller - that is making use of 24577/UDP (wally brain is awake again) - I guess trouble with discovery and management is more on that special ether-type 0x8783 for L2 than on UDP used for L3 (routed connections).

 

Some sniffing would easily proof if the switch does fail to deal with ether-type 0x8783. That's why I point on the Netgear switch engineering resp. fellow Laurent Masia. 8-)

 

Regards,

-Kurt

Message 9 of 10
PDunnington
Luminary

Re: M4300-12X12F replacing M7100-24X - Wireless Access Points not working

@schumaku

@msi

@LaurentMa

 

Kurt,

Apologies for the delay in replying.

 

Further investigation; prompted by your suggestion, did indeed point to the Switch not correctly processing/deciphering the MiNT protocol ether-type 0x8783 at Level 2. I had to change our entire AP estate over to UDP Level 3 adoption even though we are a single non-routed site (albeit quite a large one). This had the desired effect and all APs are now discovered & adopted correctly through the XSM4324S.

I still consider this as a fault/issue (which I've reiterated to Netgear Support), as no other Netgear (or HP/Dell/D-Link etc) switches we've ever had or used over the past 15 years have behaved in this way and I don't think it's unreasonable to expect a £2300/US $2800/EU 2700 switch to have consistency of features & performance with previous models in the Enterprise range.

Despite this, and thanks to your advice; I've managed to successfully circumvent this issue. Your help and advice is very much appreciated. Thanks also to everyone else who took the time to reply/comment.

 

All I need to do now is contact Netgear Support about the faulty M7100-24X that forced me to buy this XSM4324S in the first place. No doubt they'll expect me to have a Purchase order/Invoice from 2012 for it!! 

 

Kind Regards,

Paul

Message 10 of 10
Top Contributors
Discussion stats
  • 9 replies
  • 309 views
  • 4 kudos
  • 3 in conversation
Announcements