2nd attempt to use Netonix

[sirhc]

Well lets keep in mind that we are talking 120'

If you hook to batteries at the bottom and let's assume you only have 5V of drop which I think will be more under heavier loads and the batteries discharge to say 14V which means you are only getting 9V at the top and if you're up converting 9V to 50V (which is what the switch does) and then you want to deliver power for say 8 radios and 2 are AF24 and the rest are airMAX radios that is almost 200 watts peak.

How many amps will you be forcing up that little tiny wire to feed all that power?

I think Dave is being generous with the 12AWG wire, I would want at least 10AWG wire for that long of run and to be able to power all that I want to with that switch.

Your little wire is at risk of overheating and catching fire

Not to mention the voltage/current dip may result in the unit rebooting

[Dave]

sigh..Dave is never generous, only accurate....

depends on your true load, and a bunch of other items already discussed......having the voltage at the switch get anywhere near 9 Volts is just crazy because the switch might easily trigger off if it gets a low voltage glitch with a heavy load, so, best practice is always to double up whatever ya think ya need.

[wisp-wand]

2.1amps is the current load. it will be 2.4 amps once I add another radio. using the calculator Dave provided when on mains it will be around 24vdc input (due to 27 ish volt battery float). When on batteries around 21v input.

[sirhc]

2.1amps is the current load. it will be 2.4 amps once I add another radio. using the calculator Dave provided when on mains it will be around 24vdc input (due to 27 ish volt battery float). When on batteries around 21v input.

Well remember if your AC fails and you fall back to batteries your voltage will drop and your amperage will go up.

Calculate the wire capacity and tell the switch to shut down if voltage reaches a level that is too much amperage for the wire at that length

[wisp-wand]

So far so good. Pleased. 6 Hours uptime. Mind you last time initially all was well, it was after a few days it played up.

Had to disable STP as switch was blocking the Air fibres. There was no way there could have been a loop and the outgoing Cisco switch never had an issue, a bit odd.

Fingers crossed.

[sirhc]

What version of firmware are you running?

There is a reported bug that can affect STP if you have the UBNT Discovery protocol and or the Discovery Tab and a large flat segment.

v1.4.2 fixed some of the bugs and v1.4.3rc4 fixed some more.

[wisp-wand]

Chris, you recommended I use 1.4.2 and turn off discovery tab. So thats what I did. Each tower has its own management vlan, which the connected radios are in. so about 20 radios. Not sure if thats large though.

Wayne

[sirhc]

Chris, you recommended I use 1.4.2 and turn off discovery tab. So thats what I did. Each tower has its own management vlan, which the connected radios are in. so about 20 radios. Not sure if thats large though.

Wayne

Well then thats not it. I have not idea why STP is seeing a loop, possible a device with a leaky VLAN?

The STP bug from discovery (which did not affect everyone) you would see the CPU pegged at near 100% all the time.

[wisp-wand]

I will check that further, we are using Mikrotik as PPPOE and I can packet sniff, perhaps check each VLAN and see if I can work out where it might be coming from, tricky to do though as I guess I need STP enabled to see the leaky traffic, but when I do that the ports shutodwn!

The switch is still up and seems fine although CPU is 40-60% when I am logged in, is this normal?

Regards

W

[sirhc]

When you first log in the CPU is higher but if you sit there and wait it may drop.
My 24 port switches hovers around 30%+/- smaller switches usually are lower CPU.

Disable these things one at a time and see what process may be eating CPU:
1) Discovery Protocols
2) Discovery Tab
3) SMTP service

Warning: There is a fix in v1.4.3rc4 where if your SMTP server is mis-configured, the SMTP server is slow, or the SMTP server can not be reached that can mess up the switch as the SMTP alerts did not used to be a background process.

Note: CPU utilization has "nothing" to do with switch capcity as the CPU just runs the UI/CLI, stats, and Daemons such as controlling STP. All packet switching is done in the core.