Ethernet uplink speed degraded

[Dave_B]

Hello,

I have an Ap (MR46E) that has an alert of "Ethernet uplink speed degraded".

I've checked the cable by investigating the port...no errors, CRCs, drops...cable seems good.

I've bounced the port.

I have the port set for auto/auto and the port is showing 100/full.

It's firmware is fully updated.

I've set the port for 1000/full, but once I set the speed to 1000, the AP drops and the port shows not connected.

Port is showing power is still being issued with 30w.

Not sure what is happening...

This Ap has been running fine for over a year and this just started happening a week ago.

Any ideas???

[cmr]

Does the cable test show all pairs are good?  If you connect another device using the same cabling and switch port, does that come up at 1Gbps?

[Dave_B]

well I didn't do a cable test per se`...I just checked the port using cli command sh int gi0/x.

port shows clean.

I actually just noticed a few others showing the same alerts from several different sites. I have 5 locations around the globe.

[RWelch]

I had this same scenario occur yesterday with one MR36 with nothing changing in the equation.  I believe it's possibly attributed to one AP that is rarely or seldomly used.  Cables are good, firmware all up-to-date.

 

[RWelch]

I just fixed my MR36 that had a similar issue as you described above.  I turned off the port and turned it back on.  Nothing else in the equation changed.  Not sure what caused it to change to 1OOMB full duplex from 1Gig auto negotiate.  But it fixed mine.

[Dave_B]

just tried bouncing the port on 2 different APs but no joy.

[RWelch]

How about disabling the port for 60-90 seconds then enabling it?  That's what worked for me.  Hope it works for you too.

[BlakeRichardson]

Are you using Cat6, Cat5e or Cat5?  

If it's connected via structured cabling how old is that cabling?  I would suggest getting a cable installer that has a decent tester like a Fluke branded tester to have a look, it could be a crushed wire, bad resistance etc that could be causing the problem. It's not as simple as getting a link and speed negotiation.

[RichardE]

I have had this error on 5 x MR46 access points, I only upgraded to firmware MR30.7 a few weeks ago. This is the 1st time I have seen this error and the APs have been in place for years. I tried flapping the switch port on 1 AP and left it off for 2 minutes, but still have the same problem. POE on the switch port is drawing 30W class 4. 

I do not think it is a cabling issue to see it on so many APs in a short period.

[cmr]

@RichardE if they are on different networks, you could try reverting one to the previous version to see if 30.7 is the cause.  If the dashboard doesn't let you go back, support can do it for you.

[ancile]

I have the similar issue as you that happen in my environment in this couple of week, my AP is C9162 and running on 2.5Gb link speed. My c9300 multigiga switch register to meraki portal for monitoring purpose and I receive alert from meraki the link speed changed to 1Gb. I have use the cable tester Fluke to inspect from source to end and the result was pass. Need to manually shut and no shut the interface to recover.

I haven't try to fixed the speed negotiation in the switchport as to me this is not a practical solutions.

I have open TAC case with meraki TAC and the first thing he get me to do try is take down the AP and connect with a new patch cable from the switch to the AP instead of through the copper panel and monitor. 😞

[cmr]

@ancile what is the cabling standard used from switch to AP?  i.e. 1m shielded CAT6 patch lead - 27m unshielded CAT5e structured cabling - 5m shielded CAT6a patch lead

[ancile]

switch-1m unshielded CAT6A patch cable -> patch panel -> shielded CAT6A backbone -> AP

[cmr]

Hi @ancile is the solid core structured cabling terminated in an rj-45 plug at the AP end?

[ancile]

Yea, at the end of the cable terminate with shielded RJ45 modular and connect to AP

[cmr]

Sounds like at least the cable standards are all good.  You could try a shielded CAT6a patch lead, just to be safe.  Not that it should overly matter, but how long is the structured cabling run?

[ancile]

The farthest AP is less then 80meter, some AP just 20-30 meter.

[cmr]

20-30m definitely should be fine!  Have you got a good cable tester that you could put on a link?  How many APs do you have with this problem?

[RichardE]

I have the same issue with some MR46 and MR76 access points since upgrading to MR30.7 firmware weeks ago. The access points have been in place for a long time with no issues. I am going to try upgrading to MR30.7.1 just released though I did have success in shutting down the switch port for a MR76 with the issue for 3 minutes and when I open the port and the access point initialised it was back up to 1Gb full duplex connection again.

[WGPete]

I tried that with one of my APs, come back up with no error then it reappeared after about 30 minutes so keen to know if the errors re-appear for you. Such a strange issue! Waiting for Meraki to come back to me with some advice.

[WGPete]

I also started getting those errors on 16 access points acros a 52 acre site. The issue is affecting APs on different swiches in different locations, on two completely seperate physical networks. All got the error at exactly the same time (13th November at 11:01).

Strangely, some access points on the same switches do not have this error. I am unable to find a pattern. Restarted switches, access points, cycle POE ports but nothing seems to resolve it. This is the only post I have found recently and seems strange to be affecting different organisations.

[cmr]

@WGPete , did you have any firmware changes that coincide with that happening?

[WGPete]

No, I did firmware updates after the issue occured to see if that helped so rolling back those this morning and no others show over the last 30 days.

[cmr]

That is definitely odd!

[WGPete]

Yep, really is! I am actually replacing one of the switches tonight with a new Aruba switch so will be interesting to see if that fixes the issue with that affected AP.

[cmr]

Good test, what switch was in there?

[WGPete]

Currently we hjave a HP 2530 48 port switch being replaced with an Aruba 2930F. The current one isn't POE so we are using injectors for that AP.

[WGPete]

So, the access point that connects to the switch I swapped out no longer has the error on it. Whether that was due to it being off for the duration of the swap out (about 30 minutes) like some other posts have menitoned or the new switch it is hard to say but that's one less AP with the problem. I am going to try and find another affected AP that I can take down for about 30 miniutes today to see if it clears the issue.

[BananaCO]

I've encountered a similar issue although its complicated slightly by some changes within my organisation. 

We had 20 Meraki AP's running perfectly happy across a large site. We've recently closed the site and kept two of the MR46's to use in our new much smaller office suite. 

I've now setup the two MR46's with cabling direct to a Cisco Catalyst 9200 switch and they were running without issue for a couple of weeks and then on November 19th (i think) they started to complain about the link speed. I've checked the cabling, reset the AP's, forced the speed on the port. Nothing seemed to fix it.  I haven't been able to bounce the switch but I'll need to wait for an opportunity to do that.