Wireless recommendation

[Norphy]

At the risk of enraging the mods

 

We don’t use the aerohive radius...we let our smoothwall handle that...and it handles our dhcp and dns for wireless. So I can’t comment on how well..or badly the APs might do it. I’ve always rather liked the autonomous capability of the Aerohive APs to fully function in the absence of a connection to any controller/hivemanager....but perhaps that communication is part of the problem for you.

 

We don't any more either. That was one of the first tasks on my plate when I joined the organisation because the APs which were acting as RADIUS points were continuously falling over. We now have Windows handling authentication and Smoothwall handling accounting as Smoothwall RADIUS isn't granular enough

 

And we don’t use the bonjour capability either...never really thought bonjour has sufficient granular layers to control which printers appear on which devices..so we use QR codes with printer manufacturers Apps..and some rules in smoothwall to control permissions to printer vlan.

 

We use it for Airplay rather than for printers, we have AirServer and AppleTVs on the LAN so we need some kind of Bonjour gateway. The Aerohive one worked so badly we offloaded that onto Avahi instead.

 

But...I would have expected all that to work...universities probably have larger implementations...but yes possibly off load radius.

 

But ..I would still be holding the suppliers hands to the flames....because they need to earn their support costs...it’s not money for nothing...and Aerohive themselves need to back them up...and that may cost them money...especially if they don’t have competently trained staff to narrow down and identify what is going wrong for you. I assume you have raised this on Aerohive community site and making lots of noise there. Annual costs for 300 APs must be thousands of pounds...and I bet your predecessor paid for several years upfront...which you less leaverage...but you could you usefully start to make noises about taking proceedings against the company...

 

I expect it to work as well but it isn't. We are holding the supplier to account and Aerohive are being given pressure but they don't seem to care that much.

 

I assume you are hosting the controller on a VM and...you have checked it’s been given plenty of resources...?

 

That's the rub. The HiveManager isn't a controller, it's a manager. All it does is push configs down to the access points. Each access point is still its own individual entity so they work quite happily (or badly as the case may be) even if the HiveManager is turned off. But yes, the HiveManager has plenty of resources. Its CPU and memory usage is nominal, as is its disk throughput.

 

Have you tried ...turning off layer 7 stuff ..which we don’t use because it we think it’s better done with smoothwall at the perimeter..so don’t see the point of doing it at every AP?

We're not using it.

 

And we don’t use Aerohive’s captive portal either..again that’s pushed to the smoothwall perimeter....because it’s smoothwall that needs to know the user...couldn’t care much if AP knows who it is or not. [/quote[

 

The Aerohive captive portals are used on the guest and BYOD SSIDs. We use it to provide a pretty form to authenticate the guests and a AUP box for the BYOD network. The Smoothwall captive portals aren't flexible enough to achieve what we need.

 

Ah.. you mention power. We do find power demands critical. Yes, all manufacturers claim their AP is 802.11 compatible...but being compatible often means the AP is limping along...especially the 3 and 4 stream, multi Ethernet APs. We find our APs randomly reset unless we up the power on the poe switches...in fact we set ours to 802.1at...and while they don’t use 30W they must just trip over the 15W. Maybe this is somewhere to start...get some inline “at” injectors see if they help.

 

This has been one of Aerohive's and our suppliers suggestions as well. They've sent us out some injectors and so far, it's made no difference. All of our switches are 802.1at (POE+) while the APs most definitely aren't (They're AP120s, 802.11n, dual band, old and crusty). We have followed every recommendation and suggestion that both firms have made but things are still bad here. The majority of the staff have given up on the wireless because it's unreliable so when the AP120s go out of support next year, we are going to push very hard to go onto another solution.

[Roberto]

The majority of the staff have given up on the wireless because it's unreliable so when the AP120s go out of support next year, we are going to push very hard to go onto another solution.

 

I recommend Aruba. We've had very little trouble with that here ;-)

[simpsonj]

What sort of problems do you have?

 

Some areas are a bit slow, though I think that's because the Switch the Aerohive APs are connected to is 10/100 rather than 100/1000.

 

I've occasionally needed to reboot APs to make them behave, but that's all done through the web console, so not a major issue.

 

My one gripe with Aerohive is that you can't customise the Capture portal that much, and can't redirect to your own intranet page. Also, if you have the Capture portal turned on, you have to sign in each time you connect to the wireless, which is a pain for staff/student BYOD users.

[nicklowe]

Hi Norphy,

When you tested the AP120s with a PoE+ injector, where did you place this, by the switch or by the AP?

If there is an issue with the cabling causing power-related reliability issues, and to determine this one way or the other, this would need to be placed by the AP with a short patch lead and not by the switch.

 

There is an issue with the AP120 if the metal bracket is screwed up excessively tightly where it can cause a short on the back of the PCB.

 

These are both potential causes that are worth precluding while troubleshooting your issue with your partner and Aerohive.

Best,

Nick

Edited March 21, 2018 by nicklowe

[psydii]

I recommend Aruba. We've had very little trouble with that here ;-)

 

Seconded. I've only used their modern IAPs and they are awesome. (caveat - we had a very detailed RF survey and design done).

 

And if you run a serious competition they will provide a very agresive price.

 

 

I have some thoughts about a key baseline specification for any mid to large school which I'll write up here later if I get a few minutes.

[Norphy]

Hi Norphy,

When you tested the AP120s with a PoE+ injector, where did you place this, by the switch or by the AP?

If there is an issue with the cabling causing power-related reliability issues, and to determine this one way or the other, this would need to be placed by the AP with a short patch lead and not by the switch.

 

By the AP

 

There is an issue with the AP120 if the metal bracket is screwed up excessively tightly where it can cause a short on the back of the PCB.

 

These are both potential causes that are worth precluding while troubleshooting your issue with your partner and Aerohive.

Best,

Nick

 

We’ve been made aware of this one too and it has been investigated. As far as I’m aware, the brackets that we’ve got are fastened appropriately.

[AlanD]

Perhaps Aerohive have effectively given up on you...because they know that even if they now get it right...you will still leave them. From your detailed reply it looks as if you have made a concerted effort .. and much more effort than Aerohive. I can’t help think that anyone reading these forums thinking about a new wireless systems will be putting a red line through Aerohive as potential supplier. So it’s not just your supplier getting themselves a bad name...it’s all those that sell Aerohive.

 

Aerohive seems to work fine...just works....hardly ever check on it. And it clearly works well for lots of people...so they really need to be doing more to narrow down the source of your problems.

 

You describe the crashes as happening when pushing out the configuaration...does it happen at random times even after they have been working?

 

When they crash ...do they reboot? Can you still use putty to talk to them?

[Norphy]

Perhaps Aerohive have effectively given up on you...because they know that even if they now get it right...you will still leave them. From your detailed reply it looks as if you have made a concerted effort .. and much more effort than Aerohive. I can’t help think that anyone reading these forums thinking about a new wireless systems will be putting a red line through Aerohive as potential supplier. So it’s not just your supplier getting themselves a bad name...it’s all those that sell Aerohive.

 

Aerohive seems to work fine...just works....hardly ever check on it. And it clearly works well for lots of people...so they really need to be doing more to narrow down the source of your problems.

 

I'm pleased that they work for you, really I am. I have no desire to cost anyone any business and as evidenced by you, they must have customers for who the solution works. However, all I can do is share my experience of the product. We've had a lot of trouble with it which is why we're keen to see the back of it.

 

You describe the crashes as happening when pushing out the configuaration...does it happen at random times even after they have been working?

 

When they crash ...do they reboot? Can you still use putty to talk to them?

 

Sorry, you have two issues conflated there. The first issue is that these access points sometimes get a problem when we try to push a configuration to them. When we try, it fails. The only way that the access point can be recovered when this happens is to reset it back to factory settings, re-add it to the hive and push the config down again. According to Aerohive, it's because they run out of space and reseting it clears it out.

 

The access points crash pretty much at random but it's almost always during the week. We think it's due to the load being put on them. When they crash, they reboot.

[AlanD]

OK...so its probably worth a reset of ALL the access points before pushing out anything new. I assume this can be done from the hive manager (such a long time since I've used it...) That should stop them having to be factory reset - because I can completely understand that that is an enormous pain - even if its a small percentage...by the time you walk all over the school with a ladder or whatever...The trouble is when you get 300 APs updating - even a small percentage failing is really irritating.

 

You suspect the random crashes - are load related....and possibly memory related too? And there is no pattern to this - like always one of a particular set of access points (...which it might be because of load)....but I think I'd be swapping some access points around to see if the problem then goes away from that area.

 

I am conscious that AP120 units are quite old now - we still have lots though....and as I said they seem to work fine. Its the 320 and 330 that are prone to PoE issues if I forget to set higher power levels when reconfiguring or replacing a switch.

 

Can you see memory use with hivemanager? I've just looked through mine....and no AP120 is using more than 60% of memory ...but there are no APs supporting more than 30 concurrent connections....and those with the higher numbers of connection don't seem to be the ones using most memory - nor is it the ones that have been powered up for longer. And apart from the odd AP (which I know about - because either the power has been off or have been uplugged)....all mine have been working for over 13 weeks. (Wondering what happened 13 weeks ago.....perhaps there was some overnight power outage...only our core switch have UPS.... I think there is a limit of 64 connections to any AP120 anyhow....or maybe that is to one radio....

 

...I still don't understand why Aerohive are not monitoring your access points and making an effort to understand exactly why they are crashing.

[Norphy]

Can I ask what kind of switches you use? If you're using HP, what POE settings are you putting on the ports?

[nicklowe]

Hi Norphy,

 

To understand configuration push issues with Aerohive, you collect and observe the following via SSH (preferable) or the RS232 console:

 

console page 0

_debug capwap cli

debug console

 

Then, attempt to push a config and observe the output. It usually becomes apparently from that what the issue is.

 

Also check what you see from:

 

show capwap client

 

If roll back occurs because connectivity is subsequently lost because of something in the config change, this can be checked for via:

 

show config rollback

 

If there are space issues on the AP, this can be checked for by reviewing:

 

show system disk-info

 

Cheers,

 

Nick

[nicklowe]

Hi Norphy,

 

To check for high CPU or low memory, initially check:

 

show cpu detail

 

show memory detail

 

If there is a rebooting issue, initially check for the AP determined reason:

 

show system _reboot-info

 

The output of these would usually determine the next troubleshooting steps.

 

Cheers,

 

Nick

[AlanD]

It might be worth checking that your APs can do a UDP on port 12222 to the hivemanager. This might need some configuration of your switches....you check by Connecting to an AP using putty...and using command line ..

 

capwap ping [HiveManager IP address or hostname]

 

If Udp fails..it resorts to http which is not nearly so efficient...