MR42 will boot POE but not POE+ on Aruba switches...

Solved
jmguthrie
Here to help

MR42 will boot POE but not POE+ on Aruba switches...

I run HP/Aruba switches across my school system and have around 600+ MR42s and a handfull of MR52s.

 

Early on and occasionally, I would have and AP go 'bad' in that it will not boot via POE and I bring back to my office and it would work when i used the switch in my office.  Yet...the one I had put in its place in production was working which meant it couldn't be the drop/switch.  (and external power supply always seemed to work).

 

After this has happened a few times I realized something else was going on...then I stumbled upon the fact that my production switches are POE+ but the smaller switch in my office was only POE.  Lucklily I have access to a known good POE+ drop on my wall which I also use for testing, and sure enough, these 'bad' APs work fine on POE but not POE+.

 

I know i can RMA these, but collectively, I have dealt with 10 or so with this behavior and just find it odd.

 

The switch software versions on my switches of course vary from model to model, but in most cases the software is relatively current.

 

Just curious if anyone else has spotted this behavior.

1 Accepted Solution
jmguthrie
Here to help

The POE injector is for testing purposes only, I don't have any in production whatsoever. 

 

 

Its not a cabling issue as a known good AP replacing the bad one on the same run will work.  And the issue on the bad unit is replicated in a test environment.

 

And now the final piece of proof:

 

The Cisco Meraki POE+ injector test failed of course.  As of this post, I just tested with a POE injector and the unit boots fine.  There  is NO network involved, just simple YES/NO boot and the POE+ fails.

 

Hopefully a Meraki engineer will see this thread at some point.  As long as they warranty them I don't care, but I now know I am not crazy.

 

I appreciated everyone's input.  I am marking this as solved.

 

 

 

 

View solution in original post

9 Replies 9
PhilipDAth
Kind of a big deal
Kind of a big deal

I've had issues with the MR42E but not the MR42.  The MR42 has been rock solid.  Also note that while an MR42 will work on 802.3af it will retard its performance to work within that power budget.  You need 8023.at for it to run at maximum performance.

 

Have you seen if there are any firmware updates for the switches?  Does anything appear in the switch log?  Is there any chance you have exceeded the switch POE budget, hence preventing the AP from getting enough power to boot properly? For a switch that is having this issue with the AP - what does it report its PoE co assumption to be and what budget does it report it has available?

MRCUR
Kind of a big deal

This was an issue with the firmware the MR42 originally shipped with. The LLDP negotiation for power didn't work correctly so you had to manually set it on the switch port. This has been working for quite a while though. Are you on a current MR firmware build? 

 

What OS build are you running on the switches? I have a 2530 sitting on my desk running 16.05.0003 and it's negotiating PoE+ correctly with MR42's and 52's. 

MRCUR | CMNO #12
MerakiDave
Meraki Employee
Meraki Employee

Also be aware that on some switches you might need to tweak LLDP settings.  For example, there's a support doc regarding LLDP negotiation of PoE+ power levels on certain Catalyst switches here https://documentation.meraki.com/MR/Other_Topics/Low_Power_Mode_on_Cisco_Switches so just mentioning that in case there might be similar adjustable LLDP settings on your switching infrastructure.

 

PhilipDAth
Kind of a big deal
Kind of a big deal

Another thought, MR24.5 added the ability to negotiate power using CDP as well as LLDP.

 

If your switch can do CDP and LLDP - trying disabling one of them, and using just one at a time.  Perhaps the switch lacks the ability to negotiate power properly using both protocols.

jmguthrie
Here to help

More information....as for my HP switches...I have a mix of mostly 5400ZL's (POE+ modules) and 2920's.  I also have a few 2530's (poe+) and 2520's (poe).

 

So its a mixed bag, and the software revisions are therefore mixed.

 

The firmware on the MRs is 25.11 across the board and I have kept it current for the most part (although by looking its time to go to 25.12!).

 

Here is the kicker though.....i can take a known good/new AP and plug it up to replace a 'bad' one and it works, thus pointing to to AP as the culprit.  This scenario works in the field and back in my office.  There are ZERO log entries for a 'bad' AP on a POE+ switch, but on a POE, it shows normal, i.e, power applied, etc.

 

And of course, a good/new AP on a POE+ switch is normal as well.

 

In the past, I have tweaked power settings on the POE+ switches all to no avail.

 

I have even factory defaulted the AP with no luck.

PhilipDAth
Kind of a big deal
Kind of a big deal

Have you got both CDP and LLDP enabled on the switches?  If so, and one of them is not advertising the correct information then you will experience intermittent failures.

 

If both are enabled on the switch then try disabling one.

 

I think you should also try and standardise on using one firmware version on your switches.  You are getting inconsistent results and and a mix of firmware versions.

jmguthrie
Here to help

I killed CDP on my test switch and no joy.  Was a good try.  There was a time I did disable CDP on my gear.  I need to go back to doing that.

 

As for going all same firmware, I am one man with a very large network (public school system) and as much as I would love to keep the world's most perfect network, to be blunt, it ain't gonna happen.  (This is why i went Meraki on my WiFi...EASY.) I am in process of getting updated to 2920s for all of my older gear and once this is done, onto firmware via the IMC software. 

 

But you did make me remember....I have and just plugged up a Meraki POE+ power injector and alas no joy.  There is definitely something wrong on the POE+ side on these trouble units.

 

So this leads back to the switch side as again, no real consistency on model/firmware, other than a failure of maybe 2% of my overall inventory.  Not bad, but i would prefer not to deal with the hassle.  So the question is does the switch cause it or is it a failure on the AP?

 

Hmmmm......

 

 

PhilipDAth
Kind of a big deal
Kind of a big deal

The PoE injector is interesting.

 

If you plug the AP directly into the PoE injector via a patch lead does the issue still happen (perhaps there is a cabling issue)?

 

If you plug the AP directly into a PoE injector, and disable LDP and CDP does the issue still happen?  Using PoE injectors on a port that supports power negotiation via LLDP can produce unpredictable results.

jmguthrie
Here to help

The POE injector is for testing purposes only, I don't have any in production whatsoever. 

 

 

Its not a cabling issue as a known good AP replacing the bad one on the same run will work.  And the issue on the bad unit is replicated in a test environment.

 

And now the final piece of proof:

 

The Cisco Meraki POE+ injector test failed of course.  As of this post, I just tested with a POE injector and the unit boots fine.  There  is NO network involved, just simple YES/NO boot and the POE+ fails.

 

Hopefully a Meraki engineer will see this thread at some point.  As long as they warranty them I don't care, but I now know I am not crazy.

 

I appreciated everyone's input.  I am marking this as solved.

 

 

 

 

Get notified when there are additional replies to this discussion.
Welcome to the Meraki Community!
To start contributing, simply sign in with your Cisco account. If you don't yet have a Cisco account, you can sign up.
Labels