Catalyst 4006/WS-X-4013 Slow 1

[Rleeunc]

We recently replaced out 10 mb equipment with 100mb Cisco switches, etc. set uup by a VAR. Unfortunatyely the valuse added part was missing. Several of my users (including me) can't get an IP addrees at bootup. If you let the machine sit 10-15 minutes, then you can login. We're also getting scattered slowdowns in MS Outlook 98, requiring several retries for some users before they can get into their mailbox. Also, I can no longer run User manager from my PC or change my NT password. I get an error for both that the domain cannot be found. Thanks for any assistance you can provide. I'm new and drowning in this mess. I have no documentation from the VAR and very little from our previous LAN Admin who quit. RLee@NCDOI.net

 

[fynx]

Try enabling spanning-tree portfast, which allows the port to go straight to forwarding mode, not spend 90 odd seconds going through listening/learning.

NOTE: Only do this for END NODES, do NOT enable it on any port connected to another switch or router.

Cheers
Phil.

 

[martincli]

Hi the 4000 series has some funky features that are turned on by default.

Set port host a/b (where 'a' is your port and 'b' the module
This turns off trunking, etherchannel and enables spantree port fast

Also some modules have the facility to supply power to a Cisco ip phone, this needs to be turned off as well. The command is..
Set port inlinepower a/b off (where 'a' is your port and 'b' the module

This should resolve your problems.

<cut/paste>
NOTE: Only do this for END NODES, do NOT enable it on any port
connected to another switch or router.

 

[NSAI]

I have a similar issue. I have a 4006. I have solved the end user issues using the portfast on. However I am having an issue w/ bandwidth between floors. Due to budgetary reasons we have not upgraded the switches/hubs on the remote floor (coming next year!) We have a series of Netgear switches and hubs (all 10/100 auto sensing) that are uplinked (uplink ports hardcoded for full duplex)and connected to a single port on the 4006. That port is in it's own VLAN and I have portfast turned off, speed and duplexing hard set at 100/full. However I am getting ABYSMAL bandwidth. As I monitor it.. I RARELY get above 2Mbps thoroughput! My users on the floor actually get BETTER response when their NIC's are hard coded to 10/half! (whether they are on a port that is on a switch or hub!) I am sure that a Cisco 3948XL w/ gigabit GBIC connected to the 4006 would do WONDERS but that is next year's budget item. I need to get the floors communicating at full available speed until then. Any ideas or suggestions?

TIA
Stephen

 

[hibit]

Stephen,

Just a note,
Some NICs do not properly negotiate to full 100. I would set the Cisco to auto and try from the cleint to set 100 full. Also there was a problem with NT using full 100. I think this was fixed but not sure.

Also make sure you do not saturate the servers. Most clients do not need more then 10 full. Example:

10 clients at 100 full downloading from one server at 100 full == 2000mb request from clients and 200mb response from server. (You can never get this high but just for example)

Under this example one client could tie up the server with a large download and could cause TCP timeouts and retries.


Anyhow check that there is not an issue with the manufacture of the NIC.