Netgear Switches & Windows Networking

[SteveM555]

Hello,

 

We're having a bit of an odd problem. We have replaced 5 old 24 port D-LINK switches with 3 48 port GS752TS Netgear Stackable switches (connected with the SFP stacking cables). Mostly everything is working except two issues..

 

1) Devices that are directly connected to this switch will not receive an IP from DHCP. IP Camera/Printer etc. If I connect a PC and boot into Windows, it gets an IP address??? (even PXE booting doesn't work) I thought this might have been Spanning Tree/Fast Port but I've disabled this as a test and it's the same. - If I connect another switch to this and then the device, everything works fine.

 

2) It seems to have stopped the 'Windows Networking' from working. It isn't a massive issue, but with have software that deploys updates etc. via this method.

 

Everything looks fine in the switch, it has ARP entries, no errors on ports and apparently no discarded packets since last reboot...

 

Not sure what's going on?

[geezersoft]

Sounds like a possible vlan configuration issue?

[forkies]

If I remember correctly the net gear switches had a auto voice vlan and auto video vlan as vlan 2 and vlan 3. Might be this?

[SteveM555]

It does have VLans configured, but they all seem to be working fine, I will double check - All I have done is tagged the necessary ports.

 

WiFi clients on the two tagged ports are getting DHCP and are on the correct subnet though.

 

If I remember correctly the net gear switches had a auto voice vlan and auto video vlan as vlan 2 and vlan 3. Might be this?

They do, we have our VLans on 10 and 11 - These seem to be working OK.

 

If I connect another managed switch to the Netgear stack, anything connected to that works perfect, PXE and cameras etc. get an IP. Directly connected to the Netgear they don't - But Windows DHCP works. Very confusing.

Edited August 16, 2017 by SteveM555

[geezersoft]

What ip address does windows get?

[geezersoft]

What is doing the layer 3 routing and how many dhcp servers do you have?

[SteveM555]

What ip address does windows get?

 

192.168.20.x/16 - Our assigned DHCP range. Both VLans configured get 172.17.x.x and 172.18.x.x as they should.

 

What is doing the layer 3 routing and how many dhcp servers do you have?

 

1 Windows DHCP server, 2 Smoothwall DHCP server one on each of the VLans for Wifi.

 

It must be something with the Netgear switches, I re-added the old DLINK to test it and it worked fine. Must be missing something.

Edited August 16, 2017 by SteveM555

[geezersoft]

Ok have you looked at the dhcp snooping settings?

[SteveM555]

Ok have you looked at the dhcp snooping settings?

 

I left DHCP snooping as being disabled (default), as it wasn't on the previous switches. I can't understand why direct links to the switches don't work, but connecting via another managed switch (different manufacturer) works fine...

[geezersoft]

Have you disabled stp?

[SteveM555]

Have you disabled stp?

 

I'll have to give this a go Friday now, as it's going to be too busy tomorrow. If it works then at least I know the problem, but obviously the switches will need STP..

 

Failing that it's a factory reset and start from scratch. Can't find any similar issues either.

[geezersoft]

Use rstp or mstp. Stp is too slow these days.

[geezersoft]

I would be tempted to set up a mirror port on the switch and put a laptop with wireshark on that set to promiscuous mode and look at what is happening.

[AlanD]

...and you do have the latest firmware on these netgear switches?

 

We are netgear throughout - and have been for some time....but its funny you should flag up these (strange) issues as we have had a series of finger pointing issues with these GS752TS switches of which we have several. There has been a number of times when things suddenly stopped working when adding VLAN configuration to a port and there is now a known issue with using modules in ports 47 and 48 which cause port 50 to stop working completely (even when you stop using 47 and 48.. And once screwed up - I have had to reset the switch complete...put back the configuration it crashed with - and then its fine.

 

It makes me nervous every time I need to change a configuration on them. But each version of firmware does seem to get better...

[geezersoft]

AlanD has made a good call there. I have had Netgear switches occasionally misbehave until hard power cycled a couple of times (needless to say I do not use Netgear anymore).

[Sephiroth]

It does sound like it could be a VLAN issue. Have you set both VLAN membership and PVID on untagged ports? that tripped me up when I first got hold of them.

[SteveM555]

It does sound like it could be a VLAN issue. Have you set both VLAN membership and PVID on untagged ports? that tripped me up when I first got hold of them.

 

VLans seem to be working. I've tagged the needed ports in VLan membership - I haven't done anything with PVID though? In my list it is showing current PVID 1... But some are set to 0?

 

The VLans seem to be working fine though. I think I might have to wipe out the VLan config just to be sure.

[forkies]

VLans seem to be working. I've tagged the needed ports in VLan membership - I haven't done anything with PVID though? In my list it is showing current PVID 1... But some are set to 0?

 

The VLans seem to be working fine though. I think I might have to wipe out the VLan config just to be sure.

 

Yes, I had to set the pvid to the correct vlan for each end device as well as setting the port to untagged. Leaving uplinks to pvid 1. This is likely your issue.

[SteveM555]

Yes, I had to set the pvid to the correct vlan for each end device as well as setting the port to untagged. Leaving uplinks to pvid 1. This is likely your issue.

 

How does that work with multiple VLans on the same device? Our WiFi APs operate on VLAN 1, 10 and 11 to segregate the network?

[forkies]

How does that work with multiple VLans on the same device? Our WiFi APs operate on VLAN 1, 10 and 11 to segregate the network?

 

We untag and Pvid the port to the management vlan of the ap, then tag the other vlans. The ap should then route the other vlan traffic to the tagged vlans.

[SteveM555]

We untag and Pvid the port to the management vlan of the ap, then tag the other vlans. The ap should then route the other vlan traffic to the tagged vlans.

 

Thanks, I'll have a look at that.

 

On another note, I've already found one issue related to broadcast packet poisoning in the Netgear release notes (fixed in the latest firmware) "Multicast packets sent to 01:80:c2:00:00 causes a loop in properly converged STP topology",

 

I've just fired up Wireshark and can confirm Ubiquiti devices broadcast to this address. And I've just checked I don't have the latest Firmware on the switches. Going to upgrade ASAP once everyone has left.

 

I am hoping this is the problem as DHCP uses broadcast so this makes some sense with the issues we're having...

[AlanD]

Use the latest firmware...and turn off STP. Are you sure you need to use STP? I've always preferred to leave redundant cables that would provide loops disconnected...as I prefer to go around and connect one up if I lose connectivity in some outlying area and need to re-patch a route to it. STP is always capable of tying itself in knots if connections get a bit "iffy" especially when power outages (usually a over enthusiastic caretaker or cleaner) keep turning things off - especially if they turn them back on again shortly afterwards.

[SteveM555]

Unfortunately, we're probably going to have to send back all our Netgear switches I think.

 

Factory Reset the switch and upgraded firmware. Without any VLans and STP disabled, DHCP snooping disabled we're having the same issue.

 

If we plug our DC/DHCP directly into the Netgear stack, it won't ping or give out any DHCP - If we plug it into another managed switch before the Netgear stack.. It pings and we get DHCP (in Windows only) - printers, cameras, PXE won't get DHCP.

 

I've unboxed a Cisco switch, replaced Netgear as a test and everything works perfectly. Unfortunately, we need the 3 48 port switches and not just one 24 port Cisco

 

Still no idea what the issue is. I've emailed Netgear support as last hope...

[AlanD]

There is something really weird going on....because we use netgear throughout - and don't get DHCP issues.

...and we use a couple of those GS752TS in a stack.

You have upgraded the firmware in ALL the switches? And the latest firmware is the actual working version (There are two firmware images in each switch).

The DC itself has a statically assigned address? On the default or some other VLAN

And its not a stacking issue? You have tried it with a single switch?

Netgear have always been very good with support on the very few occasions I have needed to use them (a bricked switch after a firmware update..for which they were ready to send a replacement switch until I offered to try reinstalling the boot code - which they were happy to let me try - and then it worked)

[SteveM555]

There is something really weird going on....because we use netgear throughout - and don't get DHCP issues.

...and we use a couple of those GS752TS in a stack.

You have upgraded the firmware in ALL the switches? And the latest firmware is the actual working version (There are two firmware images in each switch).

The DC itself has a statically assigned address? On the default or some other VLAN

And its not a stacking issue? You have tried it with a single switch?

Netgear have always been very good with support on the very few occasions I have needed to use them (a bricked switch after a firmware update..for which they were ready to send a replacement switch until I offered to try reinstalling the boot code - which they were happy to let me try - and then it worked)

 

Hope so, just waiting for them to get back to me now.

 

Yes the firmware is on all of the switches and shows "Detected code in Flash" as the right version on all 3. Switch status shows OK in the stack. Error counts in the statistics showing 0. Haven't tried a single switch yet, that'll be next.

 

It is really odd.. apart from this everything else is working fine...