explain this one...

[mrbios]

over the past 2 weeks our networks felt very slow, today telneting onto our 4108GL i noticed a great lack in responsiveness (took about 30seconds to go down one on the menu!)

 

so from the time school finished until 5:30 i was working franticly to find the cause of this issue, unplugged cables 1 by 1 on the 4108 to find that unplugging one cable fixed it......ok simple enough i go see if theres something going on at the other end of that cable...

 

at the other end of that cable is a switch that switch wasnt at fault, but a fibre connection that went into that switch was, so i went to the other end of that, at the other end of that i did the same thing again, unplugging each connection until the problem was fixed, turned out the one i unplugged went back into ANOTHER switch in a room 2 doors down...

 

NOW, unplugging all the connections from this switch or unplugging the switch completely returns the network to normal, no extreme slow downs on the network at all, but plug just one pc in this switch and you get random spikes, plug 2 or more in and the network goes like its on oldman hamster power

 

what ive tried: tried another switch, tried each PC in the room individually, tried using different ports for certain things then it was time to go home

 

can anyone think of what might be causing this? if its not the switch and its not a particular piece of hardware then im stumped

[spc-rocket]

Hi,

 

Is the end switch has fibre connections if so it could be the fibre trancivers, these do have the habit of causing issues such as packet loss, slowness etc.

 

Check for the usual culprits:

 

= STP (spanning tree ptotocol)

= portfast enabled on all ports which will connect to workstations or servers

= Is there a loop formed when you connect the pc/devices to the switch?

 

Ash.

[tech_guy]

You sure you haven't got a loop? Sounds like it to me. Anything new done/changed the last couple of weeks?

 

Is the kit in locations where kids/staff can fiddle?

[brubakes]

We had a similar issue to what you are describing. Turned out it was a loop where a teacher had plugged a patch cable back into its own switch.

[mrphil]

do you have network diagram? though have to agree that probably a loop

[Guest richard]

We had a similar issue to what you are describing. Turned out it was a loop where a teacher had plugged a patch cable back into its own switch.

 

Same here. Sounds like a loop.

[mrbios]

We spent alot of time hunting for loops, nothing there, spanning tree is enabled too (on most anyway, the last one with the problem isnt a HP one, its an old planet one)

 

its Fibre to the first switch then copper to the one causing the problem

 

we had everything unplugged and could see the direct links between the things causing problems, having unplugged everything we can garuntee theres no loops

[maniac]

I suspect a faulty switch. Planet switchs are poor in my experience, and I have had them fail on me before. Try a different switch in the last location, If it doesn't then look at things like the cabling between it and the next switch.

 

As for the weird behaviour, I can only assume the network is being flooded with multicast or broadcast packets originating from this switch. They're hard to track down the source for sometimes, the way you've done it is often the only way.

 

Mike

[OutLawTorn]

At one school i had a PC that had frozen while a student had tried to save a document to their network drive. They simply left it and walked away, yet it seemed that in the process of freezing, it managed to cause similar symptoms as a loop. Reset the PC and it was all good.

 

It sounds like you've narrowed it down to the one switch, now you just have to see if there is any particular device or port on that switch causing the problem.

[mrphil]

when you log on the machines and in Task manager whats the network usage like on each vlan if its above 20% you probably have a loop, just turn your switches off one at a time to see when this drops down and you will isolate your problem

[mrbios]

well thats just bloody typical, restarting all the machines in that room fixed it *sigh*