WS-12-250-DC Going brain dead

[Uintadave]

Every port has its own Vlan. the MT router is powered on port 12 and has an extra on port 14.

The router then bridges the Vlans for the AP's and keeps the Links separate. This allows me to have one default config for all the switches to be interchangeable.

[sirhc]

I tried every port even the untagged ones..

I did not try the console port.

Is there a way to have the log write to something that doesn't get wiped after reboot? some way to get it to write to storage for x amount of mb?

There is a sys log under Device/Configuration but if the last thing that is happening such as MSTP, Loop Protection or such that crucial bit if information would be missing.

If you have a "spare" unit this is where I would swap the unit and see if the problem persists and if it does chances are you do not have a defective unit and there is something either on the net or environmental like your charging theory going on. If the problem stops then you know you probably have a defective unit.

[lligetfa]

Is there a way to have the log write to something that doesn't get wiped after reboot?

Setup a remote syslog server.

[sirhc]

Every port has its own Vlan. the MT router is powered on port 12 and has an extra on port 14.

The router then bridges the Vlans for the AP's and keeps the Links separate. This allows me to have one default config for all the switches to be interchangeable.

Just a suggestion, you never know when you could get hit by a falling air plane and someone has to come in and understand it maybe put a little bit more info like VLAN 12 - MT

OK so what your describing here with the whole bonding of port 12 and 14 and 14 having an SFP module I am leaning towards SFP module if you say you have this exact same config on all other switches with no issue but you say you also had issues with this model SFP before.

I must admit I know nothing about MT

shultz.jpg (54.99 KiB) Viewed 7070 times

So I am guessing you are NOT using any LAG but somehow having the MT bond the port traffic?

Another curios thing to try is remove the cable or both on ports 12/14 when on site when this happens and see if you can then gain access to the switch as maybe there is a loop?

Are you using RSTP and or Loop Protection (which is configured on the Device/Configuration Tab by a check box)? I might disable Loop Protection if it is on as maybe this is confusing things? But then again you have this "exact" setup running at other sites correct maybe just another SFP module?

[Uintadave]

First thank you for the quick replies.

Ok... the Vlans dont have the descriptions just the ports do. Each Vlan just goes to a port. The MT just puts what ports I want into a bridge. the main reason for this is that I want most of them separate.

I am leaning heavily to it being a SFP issue... with the AC switches though it would just drop the SFP port not the switch... I will try another unit and report back if it happens again or not. I will also have this one on the bench to see if it does the same thing.

[sirhc]

You should look into creating a LAG between the switch and MT instead of a bridge. I create a Static LAG between my switch and Cisco Router for all the AP VLANs and then I keep the back-hauls separate that way Pause frames generated from the AP's do not affect the back-hauls as in my topography which is an OSPF ring the traffic transgressing through one tower to another tower further out on the ring is not hindered by the local pause frames generated between the switch and router from the AP's.

Currently I configure the switch ports the back-hauls are on as a 2 port VLANs excluded from all other ports basically making a 2 port virtual switch acting as a mid-span injector but we are going to move to (2) LAGs one for back-hauls and one for local tower radios. I like feeding all the local tower radios with a 2 port LAG that way pause frames are divided out across the 2 ports in the LAG so pause frames have a 50% less impact on traffic to local tower links.

I love having all my radios including back-hauls running through the switch as I can reboot them if needed and also able to look at the graph data at a glance and know how much traffic is going where.

viewtopic.php?f=30&t=452

[Uintadave]

Interesting... I do not know how LAGs work but I will look into it.

That is one reason we just have 2 ports going to the router. One port gets all the devises that hook up at 1gbps and the other gets all the 100mbps stuff. Wouldnt that keep the pause frames from affecting the back-haul traffic?

like having all the back-haul Vlans on port 14 and all the AP's on port 12? Or would I need to separate the Tagged option in the switch?

[Uintadave]

I found the problem!!!!

It was nothing to do with the hardware. I replaced the hardware and a few days later the exact same thing happened.

What I did this time was just reboot the router and not the switch. As soon as I did that the switch started responding again.

For some reason the new config defaults RSTP to be on. For what ever reason it was conflicting with the RSTP I have inside MPLS tunnels from each site for my redundant core. Somehow it glitched and set a loop in itself. I removed the STP and enabled just basic loop protection and so far so good.

I found this out when another site did the exact same thing when a mac for an IP changed when I started doing an Airview.

[sirhc]

Glad to hear but be sure to keep firmware upgrade whenever possible as I think there were some recent RSTP fixes and you need to be aware how loop protection works.

Here is a short post on how Loop Protection works and can cause issues even if a customer creates a loop inside their network provided you are granting Layer 2 access which I suggest not doing.

viewtopic.php?f=17&t=1154&p=8644&hilit=loop+protection#p8654