Dropping ports on new WS, what is wrong with my setup?

[TheHox]

I have 2 new 24 port WS's at an MDU feeding wired and wireless access, they were working ok when we left, but about an hour or so after we left for the day they all went offline, We had to power cycle the WS to get back online.

Looks like it is doing it again now, log shows this.
port 23 I believe is the uplink to the other switch.

Code: Select all

Dec 31 21:06:27 kernel: link state changed to 'down' on port 23
Dec 31 21:06:29 STP: set port 23 to discarding
Dec 31 21:06:29 kernel: link state changed to 'up' on port 23
Dec 31 21:06:32 STP: set port 23 to learning
Dec 31 21:06:32 STP: set port 23 to forwarding
Dec 31 21:06:54 STP: set port 23 to discarding
Dec 31 21:06:54 kernel: link state changed to 'down' on port 23
Dec 31 21:06:56 STP: set port 23 to discarding
Dec 31 21:06:56 kernel: link state changed to 'up' on port 23
Dec 31 21:06:59 STP: set port 23 to learning
Dec 31 21:06:59 STP: set port 23 to forwarding
Dec 31 21:07:21 STP: set port 23 to discarding
Dec 31 21:07:21 kernel: link state changed to 'down' on port 23
Dec 31 21:07:25 STP: set port 23 to discarding
Dec 31 21:07:25 kernel: link state changed to 'up' on port 23
Dec 31 21:07:27 STP: set port 23 to learning
Dec 31 21:07:27 STP: set port 23 to forwarding
Dec 31 21:07:53 STP: set port 23 to discarding
Dec 31 21:07:53 kernel: link state changed to 'down' on port 23
Dec 31 21:07:55 STP: set port 23 to discarding
Dec 31 21:07:55 kernel: link state changed to 'up' on port 23
Dec 31 21:07:58 STP: set port 23 to learning
Dec 31 21:07:58 STP: set port 23 to forwarding
Dec 31 21:08:20 STP: set port 23 to discarding
Dec 31 21:08:20 kernel: link state changed to 'down' on port 23
Dec 31 21:08:22 STP: set port 23 to discarding
Dec 31 21:08:22 kernel: link state changed to 'up' on port 23
Dec 31 21:08:24 STP: set port 23 to learning
Dec 31 21:08:24 STP: set port 23 to forwarding
Dec 31 21:08:47 STP: set port 23 to discarding
Dec 31 21:08:47 kernel: link state changed to 'down' on port 23
Dec 31 21:08:49 STP: set port 23 to discarding
Dec 31 21:08:49 kernel: link state changed to 'up' on port 23
Dec 31 21:08:51 STP: set port 23 to learning
Dec 31 21:08:51 STP: set port 23 to forwarding
Dec 31 21:09:19 STP: set port 23 to discarding
Dec 31 21:09:19 kernel: link state changed to 'down' on port 23
Dec 31 21:09:21 STP: set port 23 to discarding

 

[sirhc]

Well you would need to provide more information

Firmware version

Post up your Config Tabs

[Rogue Broadband]

I have this issue on one of my devices that is connected at 10M-H no flow control. This is a Nethatchet that we use to monitor power levels. I have noticed that it loses connection for 2-3 seconds just like your device. I know mine is caused from TX Collisions and RX Fragments as the device is only connected in half-duplex. Mine is not a real issue as it does not affect my clients service.

Apr 21 09:17:52 kernel: link state changed to 'up' on port 10
Apr 21 13:13:50 kernel: link state changed to 'down' on port 10
Apr 21 13:13:52 kernel: link state changed to 'up' on port 10
Apr 21 23:13:49 kernel: link state changed to 'down' on port 10
Apr 21 23:13:52 kernel: link state changed to 'up' on port 10

You might look at your port details to see if you are getting either of these errors on your port. If you are and you are not connected in half-duplex you should test your cable ends or re-crimp them. Also make sure your WS is using the most current stable firmware v1.3.9 FINAL viewtopic.php?f=17&t=240

This may not be a solution to your issue, just thought I would share what I see on my WISP Switch.

[adairw]

I'm having this very same problem with switches locking up tight, no traffic no management, reboot is the only thing that fixes it. Mostly WS-12-250AC's with some WS6's connected to one.
ALL switches are running 1.3.9 and doing lots of VLAN's trunks and mid-spans. No loops in the network, this is a stand alone leg of the network right now. I can post more details later but this is a HUGE problem.

[adairw]

Chris,
Wistech is also seeing this issues. It's real and we HAVE to get to the bottom of this.
I need you to tell me what you need from us to help troubleshoot this.

[sirhc]

The only thing I know of that can lock up the switch is a power blip where the switch loses power for split second then power is restored or you have a bad unit.

Swap the unit out and see if the problem goes away.

[sirhc]

Well one other thing "if" it relates? I had a site where I ran an Ethernet cable to a business next to tower which was a BAD idea as they are 2 separate grounding systems and the cable was the bond between them. We replaced cable with fiber and problem went away.

I talk about it in this post: viewtopic.php?f=17&t=894&p=6763&hilit=+reboot+fiber+ground#p6763

I have had a switch in a storm act funny that a reboot fixed, in fact this happened twice to me this year. We tracked it back to an issue with a long buried Ethernet cable that feeds the business next door. The Ethernet cable was connecting their electric ground service to ours and the storm rolling through with heavy rain would cause a ground potential difference. Since bonding the 2 Electric services is out of the question we have decided to replace the cable with fiber which is schedule to happen soon.

Problem stopped when we injected fiber to break the bond. Is your tower ground bonded to your service ground and do you have a good grounding system at site?

[adairw]

The only thing I know of that can lock up the switch is a power blip where the switch loses power for split second then power is restored or you have a bad unit.

Swap the unit out and see if the problem goes away.

Definitely not a power blip. In our case the switch only locks up after between 6-11pm when traffic levels are at their highest.
Seems to happen once traffic approaches 60Mb.
However, I tried to force it to happen last night by running a bunch of bandwidth tests between the routers moving over 100Mb and it didn't lock up.

[sirhc]

Well all of the switches use the same design, same switch core, and same firmware.

My switches are moving upwards of 1G of traffic most of the time so I do not think it has anything to do with that.

[adairw]

Chris,
I'm well aware of that. I have a WS 12 in our data center that moves over 1.5Gbps now.
But stop living in denial. You have three users saying they having the SAME problem with your switches locking up. Two of us are associates. I suggest you take this a little more serious if you want all of us to keep buying switches.
Even if this is our problem somehow, it's reflecting on your switch.

I bet we have over 40 switches installed and working on towers, grain silos, etc and all of them have been fine until recently. Something has clearly changed. Be it a software bug that's not been detected yet OR a hardware problem in a batch. Something isn't right.