latency issue(s)

gregory.b.fell · ‎04-22-2005

This has proved most difficult to get a handle on. First I will explain our architechture.

We have our main router 'libcore', which is a Cisco 4503. We have switches plugged into this, Cisco 3548XL's. The network used to be 'fast', but we had a blip one day, and users have reported that, since then, its slow.

An example is, there is an 800 mb file on our file and print server, that some people can copy to their pc in 3 minutes (everyone used to be able to), and now, most can do in 60 minutes or so.

What is frustrating is, my pc has the same issue. However, if I bypass my netgear 1 gb switch, it can copy fast. Note, that the other users do not have a netgear switch, they go directly into the 3548xl, then to the 4503, and from there fiber to server rack.

Any help would be greatly appreciated. I know there is a lot of information I have left out that will help, so ask away. I really need to get this fixed.

smif101 · ‎04-22-2005

Are you stating that if you are directly connected to the switch that has the server on it (stay in the same VLAN), that it works just fine? That is the first thing I would check. From there I would check to ensure that the router connections to the switches are in full-duplex and that you are not taking any errors.

gregory.b.fell · ‎04-22-2005

Well, on my machine, if I plug in directly to switch it is fine. The others are plugged in directly to a different switch, and have problems.

smif101 · ‎04-22-2005

Did you check the major interfaces between the switch and routers, ensure no errors and that it is at full duplex at the highest speed.

gregory.b.fell · ‎04-22-2005

Yes, although I will again. We do have some 'hostflapping' between 2 floors, but the floor that complains the most does not have any issues there.

gregory.b.fell · ‎04-22-2005

Here are some errors on the floor that complains the most.

3w6d: %LINK-4-ERROR: FastEthernet0/45 is experiencing errors

3w6d: %LINK-3-UPDOWN: Interface FastEthernet0/39, changed state to down

3w6d: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/39, changed

state to down

3w6d: %LINK-3-UPDOWN: Interface FastEthernet0/39, changed state to up

3w6d: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/39, changed

state to up

3w6d: %LINK-3-UPDOWN: Interface FastEthernet0/39, changed state to down

3w6d: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/39, changed

state to down

3w6d: %LINK-3-UPDOWN: Interface FastEthernet0/39, changed state to up

3w6d: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/39, changed

state to up

3w6d: %LINK-4-ERROR: FastEthernet0/45 is experiencing errors

3w6d: %LINK-4-ERROR: FastEthernet0/2 is experiencing errors

khramov · ‎04-22-2005

hostflapping might be an indication of chattering nic. It might be trying to flood the swith with its ARP address.

I would also check the cable for noise and other issues. Just a pinout test won't do. I probably would need a professional cable tester. We use Fluke. It works really well. Unfortunatly, they are pricy. Good luck!

gregory.b.fell · ‎04-27-2005

I have resolved the issue. I changed the configuration of the Cisco 3548xl to 'auto' for speed and duplex mode. Not sure why it worked. The machines are still 100 mpbs and full duplex, which is what the config had in it! Any idea's why this worked?