WS-6-MINI Auto Negotiation Issue

[ncats]

We had a WS-6-MINI that was dropping the Ethernet link to the backhaul radio (UBNT PowerBeam M5 running at 1Gbps auto) on port 2. To get back up and running, we would log into the backhaul, set it to 10Mbps or 100Mbps, which allows the link to come back up. We then login to the WS-6-MINI and reboot. After setting everything back to auto negotiation, it works. Likewise, rebooting the switch on-site restores the connection. It would run fine for a day or two and randomly happen again. We eventually replaced the switch assuming it was bad and haven't seen the issue since.
We now have another WS-6-MINI that is doing the same thing at a different site entirely. This site is using a UBNT PowerBeam M5AC as it's backhaul. Also on port #2 and also negotiated at 1Gbps. We are having to reboot the switch regularly. I wish I could do more testing, but it's in production, so we are replacing it today.
I have both switches on the bench. I would like to have these replaced under warranty. However, if there is anything else I should be doing to test them, please let me know.

Firmware 1.4.6

[sirhc]

You should be running v1.4.7rc18 unless you do not want the BUGs that have been found in v1.4.6 fixed. v1.4.7rc18 has all the bugs to date reported in v1.4.6 fixed. If v1.4.7 was released tomorrow and the next day a bug was reported we would release v1.4.8rc1.

Your issue could have been many things other than the switch?

Your cable ends being unplugged and reinserted when replacing the unit could have cleaned up a bad connection?

Also what are you powering the unit with and what are you powering with the WS-6-MINI?
Could you post up the Stats, Ports, Device/Configuration, and Device/Status TABs.

Set them up on your bench and try to recreate the issue there.

You're always welcome to RMA units as we will test the units out, and in this event, leave them run for several days and see if the link speed falters but if not they would be sent back as OK.

However while they are here they would be given a thorough going over and any new hardware MODs added to the board so a complete checkup anyway.

So many times people are quick to point to our switch as the issue and most times it ends up not being the switch, here is a recent example : viewtopic.php?f=17&t=2668&p=18813#p18813

What are the MAC address of the units?

[ncats]

I will have to wait until another switch fails and try these suggestions. I'd like to test them on the bench, but honestly, It's not a good use of time. The switch looks great on paper, but if they continue to cause me problems running on the latest recommended firmware, I'll just buy something else. Hopefully you just had a bad run and I got two of them.

Being that I've already replaced two switches exhibiting the same problem and the problem has gone away, I'm still inclined to blame the switch. Especially since this has happened twice and the confirmed resolution both times was to simply replace the switch.

The first problem switch took a bit for us to finally replace it. We replaced power supplies (UBNT Airfiber24 as recommended on your site), we replaced the cables, and we tried a different PowerBeam. Replacing the switch fixed the problem. It's been running fine for months.

The second problem switch was less of a surprise. We upgraded firmware to the latest general release, but didn't try any beta firmware. Are there any fixes in the RC that would explain this? If not, why would I try the beta firmware?

So next time I'll try the beta firmware and maybe turn off auto negotiation, since it was pretty obvious auto negotiation failed on the link. I'm not going to subject my users to extended downtime though.

Sorry if I can't be a beta tester.