WS-12-250AC - Reset itself to default?

[sbyrd]

I have an issue this morning where one of my APs connected to a WS-12-250AC lost etheret link. On a remote cable test it showed the Green pair as bad and the other pairs as good. The port was running 24V .75A to a Rocket 5AC Prism Gen 2.

The port showed that is was drawing power so I assumed something went bad with the cable. This happened at around 6am. At around 7:30am when I was in the office getting replacement equipment prepped the whole WS-12-250AC went down.

It was connected via and Active LAG to my tower mikrotik. The tower mikrotik showed NO link on the Netonix facing LAG ports. Using a web power switch I remotely cycled power to the Netonix a couple of times, around 5 minutes apart. Each time the LAG ports would come up with a 1G link for a second then go back down.

Well I finally got out to the tower and found the switch had a green power light on, but non of the ethernet ports showed a link or POE enabled. The 2 ports on the Netonix in the LAG also showed no link. I checked the cable that powered the AP that went offline at 6am and when I traced it back to the Tycon surge protector I found that there was water in the line. I disconnected that cable and then turned my attention back to the switch.

As this switch fed 9 APs and serves 300 clients I decided to just replace it with a spare I bought. The spare worked fine. After I re-ran new cable up the tower to replace the one with water in it and replaced the Tycon protector I got the tower completely back online.

I brought the switch back to the office and when I put it on the bench I found it was at factory defaults. Is there some internal mechanism in the switch that could put it back to defaults without either manually doing it via Console, CLI, GUI, or the reset button? Should I be concerned about redeploying this switch somewhere else?

[Julian]

Moisture intrusion can do some pretty weird stuff sometimes. Personally, I'd send it in and have it checked out just in case, but we do provide a bench test procedure that provides a reasonable degree of functionality verification, if not reliability.

You can find the bench test procedure here: http://forum.netonix.com/viewtopic.php? ... cs+#p19221

and if you want it gone over here, the RMA procedure is here: http://forum.netonix.com/viewtopic.php? ... ure#p20516

[sirhc]

Power cycling a unit rapidly can cause any embedded device to lose its flash memory contents (config). I used to do this to old WRT routers back in the day.

You had a shorted cable which was messing with power possibly shorting out and causing power dips forcing a switch reboot.

I would NOT use Ethernet surge protectors on "passive" POE switches / devices. What they do is clamp all 8 wires together and to earth ground which is a dead short which will fry the Ethernet port in the switch.

viewtopic.php?f=17&t=3309&p=21938&hilit=+ethernet+surge+protectors#p21938
viewtopic.php?f=6&t=3267&p=21629&hilit=+ethernet+surge+protectors#p21629
viewtopic.php?f=17&t=3095&p=20773&hilit=+ethernet+surge+protectors#p20773

99% of damage to WISP equipment is from ground current not surges which Ethernet Surge protector do NOTHING to protect you from.

I would bench test that unit especially that port that got shorted.
viewtopic.php?f=6&t=2780#p19221

[sbyrd]

I did not know about NOT using lightning protectors, but it does make sense when you think about it. As I am on a commercial cell tower (CrownCastle) I expect the tower grounding system is pretty good.

I have done the bench tests and all the ports report good numbers and function properly. As the water damage in the lightning protector was not that bad my guess is the protector took most of the harm and the switch was not actually receiving a short.

The AP was reachable wirelessly, it just could not get an ethernet link. So I guess it was not a short, but probably corrosion on one of the data pairs that kept the 285ft line run from negotiating a link. The switch Cable diagnostics on that line gave all clear to every pair but pair 3/6, which it showed as DEAD.

[sirhc]

Let the unit run over night on your bench for last test

[sbyrd]

I had the switch run for 24hr powering a 24V Powerbeam on the port that had the bad cable. It was connected to a 100ft ethernet line. I then had the switch run a 48V Rocket Titanium for 3 days on the same 100ft line.

Everything looks to be running fine with no reboots, defaults, or other oddities.