Netonix Forums

Chris, I am hoping I will not put a damper on your getting more of the WS-12-xx switches out this weekend, BUT, we seem to have run into a problem with the deployment of one of them on our backhaul upgrade?

We have replaced 2 sets of links that were running Rocket M5's with Mimosa B5c's on Rocket dishes. The first set went over pretty well, and was a learning experience!
We replaced the second half a night ago (Tuesday night after midnight). The Mimosa's worked fine, but we had a problem with the WS-12-250A at the end of the circuit at our hub site.

We first checked the cable as recommended, and it showed the length as being 500ft, although I know it is physically only @200 feet. But it showed good. We then set the switch to provide 48VH on port 1. The radio came up and worked fine. The circuit came up and synced, and worked fine.

HOWEVER, when we did the speedtests from the switch, we could ONLY get @ 50 Mbps download on the 1 gig port 1. After much hair pulling (Don't have much left to begin with, and braving the high cold winds that night), we tried different setups. The switch was set to 48V, and suddenly the "download" showed @80 Mbps, lots better. We then inserted the Mimosa POE block into the picture, with it supplying the power, and the switch just being a switch with the POE OFF.

We were able to then get 189 Mbps Download, which is next to the upper limit of the 200 Mbps we are being provided at our main fiber connection!

Wanting to make sure this was not a "Fluke", we went back and re-setup the switch in the above 2 scenarios, and made sure that we got the same results.

Here is the circuit layout:
Fiber (200 Mbps)<->Ubnt ERL3<->WS-8-250A<->(Mimosa POE)<->Mimosa B5c~~~Mimosa B5c<->(48VH)WS-8-250A(48VH)<->Mimosa B5c~~~Mimosa B5c<->(Mimosa POE)<->WS-12-250A= @189 Mbps Download

Fiber --/---/----/----/---/----Mimosa B5c<->(48VH)WS-12-250A= @50 Mbps Download
Fiber --/--/---/--/---/---/--/ Mimosa B5c<->(48V) WS-12-250A = @80 Mbps Download

The WS-8-250A units are working well, the midpoint has been converted to use 48Volts DC on it. As the entire site has been converted over to DC. The WS-12-250A has been setup to use 48V DC on it by installing a hole on the back cover, but the site was has not been converted over to use DC yet, and the switch was powered by the normal AC (internal) power supply.

ALL units have been updated to 1.1.8 firmware as per your email request !

From a network perspective, I noticed that as we installed the newer equipment, from the first switch to the last. That the "Receive errors" on the circuit were pushed further down the circuit towards our hub location, where most of the older 100 Mbps equipment is connected. I would think that most of these errors are caused by "Flow Control" issues with the 1 gig to 100 Mbps transition problems. Again, I ask that the firmware allow some additional delay be provided in the "Flow Control" as a selectable option, which I think would help that situation.

The "errors" on the Network have nothing to do with the WS-12-250A issue I have first stated. I expect those errors to be eliminated as we upgrade more 100 Mbps port equipment with 1 gig port equipment.
thanks,
Wayne

wtm wrote:...
We first checked the cable as recommended, and it showed the length as being 500ft, although I know it is physically only @200 feet. But it showed good. We then set the switch to provide 48VH on port 1. The radio came up and worked fine. The circuit came up and synced, and worked fine.

Did you ever replace/re-terminate this cable? Or, test with a real cable tester?

This was done in the middle of the night, it was cold and real windy, we are looking at going back up in the day, and replacing it with new CAT6 !
At that time, we will schedule a night to do some additional testing on it. I was hoping the new "slant" feedhorns would be in shortly, and we can do it all at once ?

Didn't want to spook anyone on this, but wanted to put it out there in case if was an issue. We don't seem to have this issue on the WS-8's that are running on the system, both AC and the modified DC units!

Then again, it just might be we have a single bad unit?

I have a Mimosa B5 running on a WS-24-400A, there is no difference in the board design between the WS-12-250A/B and the WS-24-400A/B, and I am able to pull 300+ Mbps.

See this post: viewtopic.php?f=30&t=564

The other night we had a storm that took out the first radio in 4 years on one of my towers which was our MAIN airFIBER24 and sadly it also took out the WS-24-400A, well sort of.

The airFIBER 24 was completely fried, dead, non responsive.

The switch was still functioning and all other radios including 2 more airFIBERS were still running but the switch was reporting incorrect current draw on the remaining 2 airFIBER links.

Since we have a BackUP Mimosa B5 that feeds another tower in the main ring traffic switched over to the Mimosa just fine and the network remained functional and since this was midnight I decided to let the guys deal with it in the morning.

Tom showed up at the tower around 8AM and tried to use a POE brick to talk to the airFIBER with his laptop but the radio was DEAD DEAD DEAD!!!!
Now my WISP did not RMA the switch back to Netonix for a FREE replacement but they will return it for repair! I am saying this as we are small and you guys know when you suffer a lightning strike, or other acts of user mistakes such as not checking cables before you apply POE and fry the Ethernet transformers that are not covered under warranty.

Yes we will offer to repair a switch or sell you parts to fix it yourself.

Port 1 in the WS-24-400A was also DEAD but other than the wonky current readings the switch was functioning fine.

So Tom took out a spare switch and transferred the config and replaced the WS-24-400A.
Joel and Justin showed up later and replaced the airFIBER 24 and restored balance.

However traffic was all going to the internet via the BackUp Mimosa B5 from midnight until early afternoon the next day with an average load of 250-300 down and 50+ Mbps up.

Now just to prove a point that the switches will work fine with Mimosa here is a screen grab I just took. I disabled the port on the WS-24-400A on the tower that took the lightening hit the other night and failed my network over to Mimosa from the airFIBER and took a screen grab!

CLICK IMAGE BELOW TO SEE FULL SIZE

I added the second picture to make sure you know the port is Mimosa! - LOL

Also note I am pinging the internet as well.

I am located at the RF Armor warehouse right now in New Holland, PA and the Mimosa link is several airFIBER 24 hops from me and I am connectected into an airMAX M5 sector running UNI-III 20 MHz channel. This sector is shared with about 10 or 20 other clients I think, I would have to go look.

Not saying that the switch would not run the Mimosa, both WS-8's are running them fine, and the WS-12 that we had the problems with was powering the Mimosa's fine.
BUT, we could not get any decent throughput on the WS-12 when we were powering the Mimosa on it.
Port 1, setup as 48VH, only got @50 Mbps download on it.
Port 1, setup as 48V, only got @ 80 Mbps download on it.
Port 1, setup as None (using Mimosa POE brick), got @189 Mbps download on it!

Cable test showed OK, on all pairs, but showed @500 feet cable length, when we only had @200 feet of cable.

We will be replacing the cable this next week, and doing further tests on it!

wtm wrote:Not saying that the switch would not run the Mimosa, both WS-8's are running them fine, and the WS-12 that we had the problems with was powering the Mimosa's fine.
BUT, we could not get any decent throughput on the WS-12 when we were powering the Mimosa on it.
Port 1, setup as 48VH, only got @50 Mbps download on it.
Port 1, setup as 48V, only got @ 80 Mbps download on it.
Port 1, setup as None (using Mimosa POE brick), got @189 Mbps download on it!

Cable test showed OK, on all pairs, but showed @500 feet cable length, when we only had @200 feet of cable.

We will be replacing the cable this next week, and doing further tests on it!

Normally when you do a cable test and you get either pairs of different length or abnormally long lengths this indicates the cable was damaged.

However the port COULD be bad which would be easily tested by plugging the cable into another port and doing the cable test on the other port.

For instance if you plugged the same cable into port 3 and it reads the correct length and works fine then the port PROBABLY is bad and you need to RMA the switch.

If the cable length is the same on the other port and it works no differently than the cable is suspect?

The fact that you say the port works with a POE brick COULD be an indication the Ethernet transformer is bad????

I would do the simple testing I explained above to first determine if it is the port itself that is possibly is bad.

Thanks,
will do !

I am powering Mimosa with 48V and I am not having any problems. I am using 2x20 channels.

We use a lot of this cable in our own outdoor network. http://www.shireeninc.com/outdoor-shiel ... p-cable-2/