Procurve 2626 & 2650 Internet Issues

[Cragzman]

Hi all,

 

In a section of our main building, we have a cabinet with one Procurve 2626 and one Procurve 2650. Any computers or laptops that are connected to this switch are having issues in connecting to the internet via our proxy server, and pinging the servers on our network.

 

We started getting problems with the two switches yesterday afternoon and still have no idea why machines are unable too connect to the internet or ping the servers. No settings have be changed and there are no vlans set up. Switch IP addresses are static also.

 

To test, we unplugged every network cable and tested each machine individualy to see if a computer was causing the problems but we got the same lack of connection to net an all machines. But there was 1 out of 10 times where we got the net but then it goes away again.

 

I'm out of ideas and I'm now completely bald. Any ideas?

 

Cheers

[keithu]

Do your workstations use DHCP? What IP addresses are they getting? What message to you get when you try to ping your servers?

 

Presumably you've checked the web management interface of the switches for errors (?)

[Slartibartfast]

Can you ping the switches from elsewhere on the network?

Can you ping anything when connected to the misbehaving switches?

Is there anything in the switch event logs?

Can you test any fibre links to the core and the associated SFP modules?

Have you removed everything from each switch and tested them individually?

Is STP enabled on the switches?

Edited April 22, 2010 by Slartibartfast

[Cragzman]

@keithu

1. All the machines have DHCP addresses in the correct range and the correct subnet mask.

2. Pinging the servers get 4 responses of "Request timed out". After a few ping attempts, it wakes up and pings ok. But on some servers, it still don't work.

3. Yes I have checked the event logs on the switches and they don't say anything appart from the errors from few days ago.

 

@Slartibartfast

1. The switches are pingable from anywhere else on the network with no problem. But pinging is hit and miss if the machine is directly connected to the problem switches.

2. Nowt in the event logs appart from old errors.

3. How do we go about testing the fibre links and the SFP modules? We have tried unplugging the fibre with the SFP module and plugging it into the other problem switch but the same result.

4. We have remmoved all devices from problem switches and tested one by one with same result.

5. Spanning tree is enabled on all switches.

 

Cheers

[keithu]

Cany you describe your network in more detail?

How do these switches connect to the rest of the LAN?

Are your workstations in the same subnet as your servers - if not, where is routing done?

 

If there connectivity problems I'm surprised you're getting DHCP addresses reliably. What happens when you try to ping one workstation from another on the same switch?

[5tu]

Have you got vlans configured on your switches? If so, what are they using and what ip ranges are on each vlan?

 

Have you tried a hard reset on the problem switches?

[Cragzman]

All the devices on the network are on the same subnet of 255.255.252.0. The problem switches are linked together with a CAT-5 in a chain and conected by fibre from one of the switches to the center of the star topology that was previously the core switch.

 

There are no vlans on the network.

 

What we have done today:

For a test, we factory resetted the Procurve 2625 switch and inputted the settings as they should be via telnet through Hyperterminal. Made sure the fibre was plugged in and we patched in just one computer. What we did notice was that the clock on all the switches are behind by one hour but no other switches are causing issues. The computer can seem to ping a select few of the servers with no problems. But on when pinging some of the others, it takes a few ping attempts to get a response and the first response is over 1000+ms. And there is a rather small amount of unpingable servers which should be pingable.

 

Just to make sure, we are going to factory reset the Procurve 2650 switch in a bit. But pretty sure that the result will be the same.

 

Cheers

[Edu-IT]

Have you got vlans configured on your switches? If so, what are they using and what ip ranges are on each vlan?

As per the original post there are no VLANs.

[_Adam_]

To me it sounds like the link to the core switch, for it to happen on both "out of the blue" perhaps a kink or the like was introduced to the cable, interference was introduced on the run, or a number of reasons.

[Geoff]

If there is a cable issue the ProCurves should detect it. Assuming you turn the fault finder on. Is that the case with these switches?

[Cragzman]

I have just factory resetted the 2650 switch and doing a ping test from the Procurve interface isn't looking good there. Tried pinging the mail server and got 100% fail. Now thinking about hitting it with a hammer.

 

Also got the fault finders on all switches set to the highest level.

[IrritableTech]

With two workstations attached to your problematic switches, can they ping each other reliably?

[Geoff]

Can you do a 'show log' and post the last 20 lines or so for both your core and an edge switch?

[Cragzman]

Think we have fixed it. The old core switch that the problem switches linked to via fibre, had monitor port switched on for the fibre link. We switched off the monitoring and now the problem switches work as they should. No idea how or why that happened. But thanks for the help guys