Meraki Admits new M53E AP MGig functionality is broken

Solved
kevinfor
Here to help

Meraki Admits new M53E AP MGig functionality is broken

We setup a trial for several of the new MR53E MGig AP to replace some older MR32 models.. (running both MR25.10 and MR25.9 Firmware) plug them into our existing switches where the MR32 are installed ... they won't connect.. show "cycling LED lights"  move the new AP's to different switches.. same issue... plug the Switch directly into the MX Fire via RJ45 (Flat network same Subnet/Vlan) Nope....  Finally AP's directly into the MX they connect properly..

 

What Happened? Turns out we have 100% MGig & /MGig/10Gbase-T switches in our network and the MR53E has a bug with speed negotiation on MGig switches (it won't even work at 1G)  so that means the AP won't work anywhere on our network .. some might say "change your switch to force 1G" - per the IEEE 802.3bz MGig Spec switch manufactures are not allowed to do this)  so the only "solution" is to buy a "cheap" 1G POE switch & uplink it to our existing switches until Meraki decides to fix the issue..  

 

So why would a customer Buy an MR53/MR53E unit except for MGig support (you would buy the MR52) in the case of the MR53E except for the Patch antenna - also why did Meraki choose to support the slower 2.5Gbps MGig when the Cisco AP3800 Supports the faster 5Gbps - attempts to have the priority of getting this issues fixed with Support & our Sales Rep  haven't worked.. no ETA...  I guess the next step is to post it here... or contact the Publications writing reviews on the new MR53E to include this with a "Don't Buy until Fixed..." to get this issue fixed ASAP...

1 Accepted Solution
kevinfor
Here to help

Just to update this issue...

 

Meraki has released a Private Firmware fix that allows the MR53E plugged into an MGig port from a 3rd party switch to physically connect to your network & function property at MGig speeds.. effected users should contact Meraki support for the update... or IM me if they have trouble locating this fix..

 

on a separate note the MR53E is an amazing piece of hardware.. our office is 3,000 sq feet - we have a single unit with Patch Antenna (MA-ANT-3-E6) mounted on the far wall.. shooting the signal in one direction through the building with multiple walls.. and the signal strength is -34 RSSI to -67 at the other end of the building.. there is also no "omni directional" signal leak.. The performance is actually better than the Cisco AP3802 with ANT2566P4W and it can handle our density of clients with no loss of performance (we didn't need dual 5G/5G and didn't want to pay for things like Prime Infrastructure that's included with Meraki..  

View solution in original post

6 Replies 6
Uberseehandel
Kind of a big deal

You are aware that Wi-Fi is half duplex . . . so, in the real world, the AP should work fine on dual 1000Mb Duplex links in most deployments.

 

"Turns out we have 100% MGig & /MGig/10Gbase-T switches in our network"

 

and how many devices that attach to the network can make use of those connections? Not the Wi-Fi ones, that's for sure. We are all going to need 1000Mb access at the network edge, for some time to come.

 

 

 

Robin St.Clair | Principal, Caithness Analytics | @uberseehandel
PhilipDAth
Kind of a big deal
Kind of a big deal

That is pretty stink about the MGig compatibility issue.  What brand of MGig ports are you using?  You are very lucky to have all MGig ports.

 

I agree with @Uberseehandel - it's not possible to drag out more than 2.5Gb/s of data throughput from an MR53.  Supporting a higher connection speed would only limit the maximum cable length with no other benefit.

 

The other kicker that also limits the throughput of the MR53 is that most clients are 2x2.  You get the occasional 3x3 client.  To date, their are no 4x4 clients.  So you will never even get a high bandwidth client to really max it out.

MerakiDave
Meraki Employee
Meraki Employee

Sorry to hear there might be a known issue with multigigabit negotiation on certain switches, I'll check if I can find a status on it, but I'm sure it's being worked on.  It's not "if Meraki decides to fix it" but what is the reasonable time to do an Engineering reproduction, root cause analysis, and write and deploy a firmware fix.  Meraki has a good history of fixing similar bugs in record time compared to other vendors because of the power of cloud management... being able to see (on the back-end and on a world-wide scale) exactly how many products/customers are impacted, generate appropriate bug reports, pop Dashboard banners if applicable, and roll out firmware fixes in a methodical & controlled fashion.

 

To the 2.5Gbps point on the MR53, @PhilipDAth is correct and I'll add that Meraki did not arbitrarily choose the lower of two data rates and select 2.5Gbps versus 5Gbps. It's physics. The 3800 has a software-defined XOR radio and can do dual 5GHz, and a 5Gbps mGig rate is warranted/required.  The MR53/53E/84 have one fixed 2.4GHz and one fixed 5GHz radio.  So roughly speaking, running all 4 spatial streams (clients can't do this yet) at 80MHz channels with MCS9 (256QAM, 5/6 coding) and short 400ns guard interval, the theoretical max is 433.3 Mbps per SS, so roughly 1.7Gbps on the 5G radio.  Add another theoretical max of 800Mbps on the 2.4GHz radio and there you've got a rather convenient AP max data rate of 2.5Gbps. 

 

 

Even if we had 4x4 clients in the mainstream, the only way you'd really stress the AP to that level (today) is probably in a lab environment, and only with the most pristine RF conditions.  Very unlikely in any typical deployment.  Now you might be thinking what about 160MHz channels, fair point, as that takes us closer to 3.5Gbps theoretical for the 5GHz radio alone, and perhaps more of an argument for 5Gbps mGig on MR53/84.  I'd argue that's also a rather large unknown.  It will certainly be some time before enough spectrum is allocated to have a reasonable channel plan for the enterprise with 160MHz channels.  And even longer before client devices with 4x4 and 160MHz channels go mainstream.  By that time, we likely won't be designing with MR53/84 anymore and will be into next-gen 11ax APs.

kevinfor
Here to help

the switch models are all Netgear   XS724EM (all 24 ports are 10Gbase-T /MGig) we mainly use 10G for server & Storage links using LAG (Port bonding) -  M4200-10MG-POE+ (All Ports are 2.5 a few 5G ports)  MS510TXPP ( 4 -1G  the rest 2.5/5G and 10Gbase-T & SFP+ Uplink) unfortunately Meraki doesn't make a 10Gbase-T/MGig switch - the MS350-24X is a 1G with a few MGig ports.. the M425-16 is 10G SFP+ not 10G copper)

 

The real issue in the case is not "1G vs 2.5G connectivity on the port & does it make a difference - but rather the AP's can't even connect when plugged into an MGig port that is forced to 1G speed (we are still working with Netgear to understand why the 1G ports on the MS510TXPP work with the MR53E but a 2.5G port forced to 1G on the M4200 or any other cannot) to be fair.. the switch "should" be able to trick any device it's plugged into to think it's a standard 1G device.. regardless of any issues Meraki has with MGig on 3rd party switches..

 

The purpose of posting this to the community was because Sales said ... talk to Support..  Support said "I have attached your case to the Mule.. no ETA on fix..."  now that the community is aware it raises the severity of the issue with potentially more customers opening cases on the issue to raise the priority of getting it fixed.

MerakiDave
Meraki Employee
Meraki Employee

Understood, thanks @kevinfor for the details and yes, I'd agree that posting to the community could prompt additional customers to try some quick tests, reproduce the issue, and log a case with Meraki Support.  Good thought.  That can only be a good thing as it gives Support and Engineering additional data points to compare with their own reproduction.  I know that frustration when Support escalates something to Engineering and the only status is "no ETA" but you can certainly ping Support for weekly updates perhaps, chances are they might be able to see some internal notes/progress they can report on.

kevinfor
Here to help

Just to update this issue...

 

Meraki has released a Private Firmware fix that allows the MR53E plugged into an MGig port from a 3rd party switch to physically connect to your network & function property at MGig speeds.. effected users should contact Meraki support for the update... or IM me if they have trouble locating this fix..

 

on a separate note the MR53E is an amazing piece of hardware.. our office is 3,000 sq feet - we have a single unit with Patch Antenna (MA-ANT-3-E6) mounted on the far wall.. shooting the signal in one direction through the building with multiple walls.. and the signal strength is -34 RSSI to -67 at the other end of the building.. there is also no "omni directional" signal leak.. The performance is actually better than the Cisco AP3802 with ANT2566P4W and it can handle our density of clients with no loss of performance (we didn't need dual 5G/5G and didn't want to pay for things like Prime Infrastructure that's included with Meraki..  

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