Wi-fi extremely slow to broken since returning from Summer - but BYOD wifi fine....

[kennysarmy]

As the title : Wi-fi extremely slow to broken since returning from Summer - but BYOD wifi fine....

 

Details:

 

MERU MC3200 controller connected to core switch model : HP Switch E5406zl(J8697A)

 

In essence we have two main wifi's. LAN & INTERNET

 

INTERNET is for BYOD and staff iPads and Apple TVs - there are no problems reported on any of these and all are working fine.

LAN is for our school owned laptops which are on a separate vlan and we've laptops in areas such as Science, DT, Library and SEN.

 

Taking one particular area of the network where AP's are connected, Area B - laptops are either taking up to 7 minutes to logon or are failing to logon.

 

I've taken a test laptop to this area and plugged it directly in to the edge switch (Switch 2520G-24-PoE) and I get good network bandwidth back to a PC back in my office.

 

Connection is laptop -> 2520G-24-PoE edge switch -> trunked fibre connection to core switch E5406zl -> UTP connected edge switch -> PC

I was getting 350 Mbits/sec.

 

When I then pulled out the network connection to the 2520G-24-PoE edge switch and enabled the wifi card in the laptop it immediately picked up the LAN wifi - but I was unable to ping anything or run another bandwidth test (using iperf)

 

I recreated this test in my office, where the fixed connection network speeds were slower ~ 30 Mbits per second, but when switching to wi-fi I was able to get the same results and able to ping other servers.

 

My office potentially is the only area in the school where the LAN wifi is working - and the edge switch I'm connected to is in turn connected to the core switch via patch leads - all other edge switches are connected to the core via SFP modules - fibre links - SFP modules.

 

 

I've rebooted the Meru controller in case it was having "issues".

 

I've rebooted edge switches and AP's.

 

 

I just can't get to the bottom of why it works in my office and no where else - and why it's occurred in the first place when we've had years of no issues with school owned laptops connecting successfully on the wi-fi.

 

Any helps or ideas to help troubleshoot gratefully appreciated.

 

To add we made no changes to any config's on any switches over the summer and no work was done to the network or any servers that might be affecting logon speeds.

 

The test laptop we are using is on the domain, but we're using a test user with no policies being applied to it and no logon scripts.

[psydii]

Are the two ssid on the same channel? If yes, you can probably rule out interference, and possibly broadcast flooding.. Though Meru might be guaranteeing airtime to the essid that isn't flooding... Not sure how clever it really is.

 

If the are, you may have a problem on the channel affected.

 

Do you get a DHCP IP on each network? Can you ping to and from it? Does the DHCP server believe it has issued an address?

 

What does wireshark show on the channel? (If you can get hold

Of a mac book pro you can get better insight from wireshark on wireless than is possible on windows due to the was the drivers work)

 

Also check spanning tree.

[kennysarmy]

Are the two ssid on the same channel? If yes, you can probably rule out interference, and possibly broadcast flooding.. Though Meru might be guaranteeing airtime to the essid that isn't flooding... Not sure how clever it really is.

 

If the are, you may have a problem on the channel affected.

 

Do you get a DHCP IP on each network? Can you ping to and from it? Does the DHCP server believe it has issued an address?

 

What does wireshark show on the channel? (If you can get hold

Of a mac book pro you can get better insight from wireshark on wireless than is possible on windows due to the was the drivers work)

 

Also check spanning tree.

 

Hi;

 

Yes the two SSID share channels, but always have done:

 

802.11bgn is broadcasting on ch 1

 

802.11an is broadcasting on ch 36

 

All AP's are set this way....

 

 

Yes, DHCP is working, as when we logon locally to a laptop and disable, re-enable the wireless network card is automatically connects to LAN and gives a good signal strength and windows reports good speeds between the laptop and the AP.

However, in my office when we do this test we can PING other resources, but in all edge switch areas where they are connected back to the core via fibre I cannot ping other resources.

 

DHCP server looks to be fine.

 

 

Perhaps wireshark is the next step to compare traffic from a working and non-working area.

[kennysarmy]

Rebooted the core switch.....all edge switches re-connected (phew)

 

The technician has done some more testing this morning, using a laptop which is in it's normal OU for the SEN dept. (all computer policies applying fine)

 

He used a newly created "slowlogon" account (domain admin but no logon scripts and in an OU where no policies are set) and the logon times to desktop appearing were:

 

Our office : 10 seconds.

 

3A1 : 10 seconds

 

3A2 : 35 seconds

 

6th form common room : 13 seconds

 

Hall : 14 seconds

 

Reception : 26 seconds

 

2A3 : 1 minute 25 seconds

 

He also tried using our normal domain admin account : SADMIN

 

In the Hall it was 48 seconds where previously in all prior tests he could not logon at all...

 

However in reception it was 6 minutes 15 seconds and in 2A3 it was 7 minutes 8 seconds.

[kennysarmy]

Update:

 

Test laptop in OU with no policies and inheritance turned off.

 

Test accounts for logging on tests; slowlogon (domain admin with no home drive or login script) in a test OU with no policies and inheritance turned off & sadmin (our normal domain admin account)

 

Area 1 : slowlogon logged in in 2 seconds

Area 1 : sadmin logged in in average 5 minutes

 

Area 2 : slowlogon logged in in 2 seconds

Area 2 : sadmin logged in in average of 20 seconds.

 

 

 

Area 1 is an edge switch connected to the core via fibre, ports reporting connection at 1000 Meg

 

Area 2 is an edge switch connected to the core via patch lead, ports reporting connection at 1000 Meg.

 

 

Network speed of switch in Area 1 back to PC connected to core good and all other LAN traffic logons for desktop PC's are good!

 

 

Am I right in thinking this MUST be a hardware issue?

 

If it was a rogue group policy or global setting it would be affecting Area 2 as well are Area 1?

 

 

 

I have many Areas which exhibit the same issue as Area 1 - the commonality is they are all connected to the core switch via fibre / SFP modules.

Edge switches vary in make and model, some AT switches, some HP, some with PoE, some without.....

[kennysarmy]

Latest test:

 

Took a spare switch and spare SFP modules and fibre fly lead and connected this all up next to the core switch.

 

Set port A14 on the core with settings as per the local edge switches for VLAN etc.

 

Moved our office AP from the patch lead connected edge switch to this new SFP connected edge switch and everything still works fine in my office.

 

I wanted to do this as it seems we are having problems in all edge areas connected back to the core switch via fibre/SFP modules, but the local edge switch connected to the core via patch leads is the only area where it all works fine.

 

This test did n't seem to prove anything as the logon times are still quick with either set-up.

 

 

We are currently trying the following test:

 

Logon locally as "administrator" - disable wireless network card - enable wireless network - should immediately connect to LAN wifi without issue and within a few seconds - it does this fine in my office, however it appears not to do this in other areas around the network - I'm just going to gather some screenshots of what happens at other locations.

 

I think this is closing in on the problem, as without the laptop connecting properly to the network when the laptops is turned on the next part of the operation "logging on" will fail or take many minutes...

 

What could be causing the laptops to fail connecting to LAN in other areas?

 

Could this be a RADIUS problem?

[featured_spectre]

are there any networking loops at all?

[kennysarmy]

are there any networking loops at all?

 

Do you mean on the core switch?

 

ie. a patch lead plugged in at both ends across the switch?

[kennysarmy]

This is now fixed....

 

A new edge switch was put in in the summer and for some reason I'd set the IP Address against the LAN vLAN and given it the IP of the core switch (?) - instead of this being disabled....I can only guess as I was setting this switch up ages ago before installation that I had my concentration disturbed