Interesting Issue with 10M port

[cbl]

Upgraded to 1.3.3 on a few switches last week. Saturday, it appears, one of my UBNT APs dropped to 10M, which sometimes happens with cabling, etc..not really the issue here.
Since that time it dropped my smokepings to OTHER devices on that same 12 port netonix started going crazy like they were also at 10M even though they were negotiated at 100M(FC). Now these other two devices (SAF Freemiles) are on DIFFERENT MGMT vlans for their pings than the AP and the netonix itself...but they DID share a common customer data VLAN.

I have yet to sit down to reproduce this, but once I forced the AP and Netonix to 100M on both sides, pings to the SAF Freemiles calmed down and have gone back to normal (see attached).

Wondering if something is getting leaked somewhere?

Again, I'll sit down and try and reproduce this. It was just weird that two different devices (each FreeMile is on its own MGMT VLAN) each on a different MGMT vlan and IP network were impacted by this latency.

[itwnz]

Anyone else seeing this?
I replaced an 8 port Toughswitch with a WS-12-250-DC last week and have had reports of VoIP calls breaking up.
Checked latency across the network and I'm seeing random spikes to the switch and anything beyond.
If logged into the Netonix web interface, I get the popup about lost communication when these spikes happen.

I've since noticed that one of our Ubiquiti AP's has dropped to 10Mbps and it's when traffic is put across this link that latency on the switch and everything connected to it goes to shit.
Cabling seems fine and the Ubiquiti had been running at 100Mb for over a year on the Toughswitch.

Currently running 1.3.7 with i2c errors - will try 1.3.9rc4 tonight.

[sirhc]

@itmz
v1.3.7 and you installed last week?

If you installed last week then that meant that v1.3.8 was already out there and the release notes tell you there was an I2C issue repaired for the DC switches in v1.3.8.

The piece of paper that came on top of the switch asks you to upgrade the firmware to current version BEFORE you do anything with the unit.

FIXES
- Select all ports and add POE did not work correctly
- Port 1 Enables itself on reboot if disabled
- UI formatting Bugs
- Changed PoE Smart to now only check for "cross-shorts"
- Change voltage calculations to use 1 decimal point of precision
- Fan failure detection changes
- Better I2C error handling, primarily helps DC switches but better for all
- Fixes for DC switches Power Tab conditions

ENHANCEMENTS
- Add configurable config save timeout

KNOWN ISSUES
- Ubiquiti Discovery Protocol Still Broken
- large flat networks MAC table requiring too much CPU time
- IPERF is still a work in progress

If you are having issues with speed/duplex then run cable diagnostics on the ports in question. Click the little gear to the far right on each port row on either the Status or Ports Tab and select Cable Diagnostics. You can run cable diagnostic with and without POE power applied.

With the SAF Free miles I hope you are using 48V NOT 48VH?

Also post up screen captures of your Config Tabs (Ports/VLAN/Device) and any Tabs that you have made changes too. This might be your first WS but there are over 7,000 switches out there in service all running fine so it has to be something in the setup and we need to figure this out.

I am available today if you have TeamView to look at this so PM me your cell numbers if you want me to take a peak.

[itwnz]

I prepped it to install before last week and used the latest firmware at the time, wasn't aware of i2c problems until voltages disappeared.
Anyway, the common theme between myself and cbl's post is a port dropping negotiation to 10Mb Full and then any traffic on that port causing havoc with high latency across the rest of the switch (or at least the backhaul link).

After rebooting the Nanostation the link is back at 100Mb and latency is fine when traffic is pushed across.

There's nothing out of the ordinary with this config, 8x Ubiquiti M series radios on ports 5-12 set to 24V - all 100Mbps devices.
1x Webcam at 100Mbps and a solar charger at 10Mb/Half.
Default settings on ports (flow control is enabled for all ports but icon is only showing on powered ports)
SNMP/Discovery turned off
Single default VLAN.

Could there be an issue with Flow Control feeding to/from a 10Mb port and 100Mb backhaul?

[sirhc]

I prepped it to install before last week and used the latest firmware at the time, wasn't aware of i2c problems until voltages disappeared.
Anyway, the common theme between myself and cbl's post is a port dropping negotiation to 10Mb Full and then any traffic on that port causing havoc with high latency across the rest of the switch (or at least the backhaul link).

After rebooting the Nanostation the link is back at 100Mb and latency is fine when traffic is pushed across.

There's nothing out of the ordinary with this config, 8x Ubiquiti M series radios on ports 5-12 set to 24V - all 100Mbps devices.
1x Webcam at 100Mbps and a solar charger at 10Mb/Half.
Default settings on ports (flow control is enabled for all ports but icon is only showing on powered ports)
SNMP/Discovery turned off
Single default VLAN.

Could there be an issue with Flow Control feeding to/from a 10Mb port and 100Mb backhaul?

No not a flow control issue but you can easily test that by turning it off.

The reason you do not see the FC icon on the Status page for some ports is that icon indicates Flow Control is negotiated and active.

Post by cbl is old, back from Oct 12, not sure he did not already find his issue?

I would try replacing the NanoStation in question, if that does not help try using a different port in the switch and see if problem follows device to new port or goes away.