Packet Loss, Negotiation Issues

[kshogan]

We re running 1.2.2 on the switch. We have a B5, a couple Power beams, and two Rockets.

The switch constantly is losing packets. We've swapped every cable, even radios, and the results are the same.

Installing a dumb switch and POEs works flawlessly. We never see the B5 of new AC PBs negotiate up to Gig, and have to hardcode negotions.
Help

[sirhc]

Well it is "possible" the switch is bad but I highly doubt it as I have never seem "all" ports mess up but it is possible?

I have seen a bad port as in singular.

Do you have Ethernet surge protectors installed as they could be your issue, try removing them and see what happens if you do.

There are over 2,000 of these switches in service and the consensus is they improve their network performance a LOT so I do not think there is anything wrong with the design of the switches or the firmware.

Can you post a screen shots of your Port Tab and screen grabs of Cable diagnostics and port statistics of the ports you are seeing the errors on.

[kshogan]

Just tried to run a cable diagnostics on the switches primary feed, (B5 Mimosa) and it hung the switch. Won't ping. B5s Ping fine. Running 48VH on that port.

Now we have to trek to the tower (2.5 hours) and power cycle it. Good grief....

We have about 150 Tough switches deployed, and the design of these Netonix's would allow us fewer "TS-Carrier" versions that are just 2 tough switches in a 1U rack. But it appears that our initial deployment has been a large failure. We understand the desire of companies to ship hardware with the thought of "fix the firmware in the field" but a switch is a critical component of a network, and running remote diagnostics should not hang it, or a laptop should not take 20-30 seconds to negotiate. It should not drop packets, ever, ever, ever. Negotiation should work on radios like B5's, UBNT AirFibers, and PowerBeam AC's.

Are these ready for production, or should they only be deployed in test networks for now? We assumed that we could use them for production, but are paying a steep price for doing so.

If there are any known issues with negotiation or packet loss, it should be clearly disclosed to customers so we can discern using the product and the risk of doing so. Are you claming that there are no known issues with packet loss and negotiation issues and this is unique to our setup? If so, how do we get a replacement switch?

Are there any options to resolve this?

[WisTech]

I realize how frustrating something like this is, but understand that many of us have a dozen or more of these things (between models) deployed without issues. With any type of manufacturing, while unlikely, it is always possible you got a bad one because I have various models all running either 1.1.8 and 1.2.2 and do not have problems using them as midspans for AF5's, or Airprism APs, etc. along with full per port VLANs and powering an Exalt 11Ghz unit at one location. If the switch is legitimately 'bad', Chris is a stand up guy and I'm sure get you a replacement. He wants every one of his customers happy and will provide you with a working switch if yours in fact has a manufacturing defect.

[kshogan]

What is frustrating is that we can't get a pinpoint of the issue. Plugging in the same cable, to the Mimosa POE, to a MacBook Pro, and negotiation happens instantly, at 1Gb, with no packet loss. As soon as the switch is added, it sometimes chooses 100M/Half, or full, then 1G etc.

Changing ports yields the same. We have several of these, as we ordered them to replace the TS-Carriers in the tower racks. We will take another TS back up and replace. If it works, and an identical configuration Netonix does the same, then at least we know where the issue is.

[sirhc]

What is frustrating is that we can't get a pinpoint of the issue. Plugging in the same cable, to the Mimosa POE, to a MacBook Pro, and negotiation happens instantly, at 1Gb, with no packet loss. As soon as the switch is added, it sometimes chooses 100M/Half, or full, then 1G etc.

Changing ports yields the same. We have several of these, as we ordered them to replace the TS-Carriers in the tower racks. We will take another TS back up and replace. If it works, and an identical configuration Netonix does the same, then at least we know where the issue is.

I am not sure what your saying, if you take another one and it yield the same issue then I have to say there is something you are doing in your config and/or setup that is causing the issue as there are a TON of people including me using these in the field with no issues.

HERE IS A POST SHOWING A SWITCH POWERING AN MIMOSA M5, 2 AF24, AND A ABOUT 12 airMAX RADIOS WITH NO ISSUES
viewtopic.php?f=17&t=742&p=5487&hilit=MIMOSA#p5487

There are over 2000 in the field with only 13 RMAs and of the 13 5 were we just upgraded the firmware as the user did not upgrade them and then ship them back and their problems went away.

I am all for helping people like you try and figure out what is wrong but it starts with providing some information that helps.

Model and Firmware

General configuration which can be provided by print screening and posting your Tabs.

Are you using MSTP, VLANs and such.

Give us a general make up of your network and how this tower is setup.

Check the log to insure no loops are going on.

But there are MANY people powering MIMOSA and airMAX radios with NO problems.

As long as your firmware is at least v1.1.8, preferably v1.2.2.

As far as developing the firmware in the field we were developing the firmware since March 2014. It went into service at my WISP in August of 2014 and we released to public on October 2014.

We released soo many firmware versions not just because of bugs but people kept asking one more feature.

There was 2 major bugs fixed in v1.1.8 but since v1.1.8 the firmware is SOLID.

Gather some information and please provide it here so we can help.

If you like you can PM me your cell phone and I will call you as I have been working on the RF Armor web store all weekend anyways, got to sleep at 5AM :(

[kshogan]

Model and Firmware
---- 1.2.2 WS-12-250A

General configuration which can be provided by print screening and posting your Tabs.

---- None other than turning on power on ports, and FC on port 1 to try and fix all errors and dropped packets, RSTP enable.

Are you using MSTP, VLANs and such.

-----No

Give us a general make up of your network and how this tower is setup.

Fiber ----{B5}--<=====>---{B5}------{WS-12-150A//////{ 4 Other Radios - 2 Powerbeams, 1 Rocket, 1 Powebeam AC }

Check the log to insure no loops are going on.

Nothing other than up/down on port 1. No loops reported.

Thank you for any help. We plan to head up to the tower and replace the WS-12 with WS-16, to see if it stabilizes things. If not, place the toughswitch back into operation.

[sirhc]

From what you're saying that is pretty straight forward ass simple config.

It could be a bad switch I can not say but there is nothing from what you say that would be an issue.

Either it is a bad switch or there is something here we are not seeing?

Let us know.

[kshogan]

We went to the tower today.

We replaced the Netonix with the old ToughSwitch Pro. All radios negotiated correctly and duplex was correct. No packet loss between devices, so there must be something wrong with this particular switch. We disabled the B5s from the switch as well. Screen below.

At one point, the WS-12-240 was showing a Powerbeam at 10M, and when unplugging and placing the exact cable into the TS-Pro, it trained up to 100M/FD Negotiations were all over, and link coming and going on different ports.

We did not have another 12-240, only a 16 port, so we did not change. We'd like to RMA this switch, and then move the configuration over and try again.