CAP ax Ethernet port speed changing, dropping off network

Hardware MikroTik hardware general
1

I have two CAP ax’s running 7.15.2, daisy chained off a Netgear GS728TPv2 which is providing POE (combined draw 54v/13.8w, netgear has plenty of power remaining as all my cameras are yet to be installed), and they are not working well.

The daisy chain setup is temporary until the electricians come to do the low voltage pulls so both can be home runs (“soon” is what I hear).

The one in the center keeps changing ethernet speed on the ethernet port facing the netgear. The netgear mostly accommodates it, but I think at times does not as this morning the CAP was off the network, the netgear showed 10mbs.

I have swapped the ethernet cable itself without change. It is 100’, literally taped to my ceiling at the moment. Prior was about 50’ before swap. The secondary cable run is about 40’. All look like 5e cable but I didn’t see a label, but I’ve used these frequently in the past at 1g for testing. Cables test good (continuity tester only, not frequency test, I don’t have that).

I am logging to a syslog server but there is nothing I see relevant in the log; it does show the speed change but nothing about why it changed, just up at 100mbs (I never saw in the log 10mbs, I think that one disconnected it).

I just swapped positions of the two CAP’s, to try to see if it is specific to one of them; time will tell.

The advertised speeds for both devices are correct. After a power cycle there is no hesitation, negotiation completes at 1g full every time.

I’ve searched for similar issues without useful results (but confess to not being that familiar with where to look in the mikrotik ecosystem).

Questions:

  • Are there any known issues with auto-negotiation on these, either in general or with other vendors like netgear? I can change to no negotiation but wanted to understand the issue first.


  • Is there any reason the daisy-chaining of these could cause it?


  • Does this sort of speed change match any known hardware failure modes in this hardware?


  • I started on 7.15.1 if I recall, and this seems worse with 7.15.2, though not enough data to say for sure. Any known issues?

My plan is to see if the problem moves to the other end having swapped them, and go from there, probably with a hard coded rate. But I welcome the collective wisdom of experienced users.

Linwood

2

Well I’ll assume your using at least 24 or at worst 26AWG cat5/6 for such a long run.
Cap ax’s are power hungry beasts.


Powering
Details
Number of DC inputs 2 (DC jack, PoE-IN)
DC jack input Voltage 18-57 V
Max power consumption 40 W (most likely CPU load related )
Max power consumption without attachments 11 W
Cooling type Passive
PoE in 802.3af/at
PoE in input Voltage 18-57 V


What does system->health look like on the units ?

3

I have a bit more. The electricians showed up and did my low voltage pulls today. Before that, with the daisy chain order swapped, no failures.

About 8pm or so I finished punching everything down and have nice shiny new cat 6, 23awg wire run to each, shorter runs, separate home runs.

A bit after 10pm the same one failed the same way – the switch went to 10mbs, the AP – no idea, couldn’t reach it. I forced the switch to 1g/full - still couldn’t reach the AP. Power cycle the port, came back up at 1g, I have now hard coded the pot to not negotiate just run at 1g.

The other CAP, and the two hAP’s I’ve got online (but not POE) have not had this problem.

I am monitoring it with zabbix which continually tracks a lot of stuff, like temp and cpu. It’s basically loafing, only half a dozen registrations mostly IOT, it is not set up for any routing, etc. The two CAP’s are each drawing a bit less than 7 watts (6600 and 6300), and no POE errors. Temp is 65C (this one is in AC space, the one that is working fine is in a garage and runs about 75C during the day, without issue).

Neither switch nor CAP show any dropped packets, which is interesting because when I look at my monitoring, that CAP and only that CAP (also including hAP’s) is showing significant ping loss (I send 10 fast pings every 60 seconds). Over the last 30 minutes (roughly since I reset it) I’ve lost 7.6% of those pings with the biggest interval being 4 of 10 loss. These are going over ethernet not wireless.

Ah… just noticed this (log from the CAP – interestingly it’s not showing in my syslog, it must dump the “down” syslog message because ethernet is down, and not buffer it). The switch shows nothing in its log.

I think this CAP is just defective. I guess I need to figure out how to get warranty. It’s like 3 days past the Amazon return period. :frowning:

Unless anyone has better ideas.
log.jpg

4

I realized I was not sure after the rewire if I was on the same port, so I switch ports to another one. This time I could not even connect. It would draw power, bring the port up (hard coded to 1g/full) but couldn’t ping it at all, then the port went down.

So… dead. Or a reasonable facsimile.