Intermittent Loss of Connectivity!

[RatherBeSkiing]

I have a problem with loss of connectivity to a new Dell Poweredge 2850 server.

The situation is as follows, I recently started a new job and have purchased some new equipment, 2x 3Com Baseline 2226 switches and 1x Dell Poweredge 2850 server. There are existing servers in place some running NT and some W2k, the new server is running 2003 with AD.

If I run a continuous ping from one of the old servers to the new server 95% of the time it is fine, but there will be 1-2 minute periods when the ping request times out? At the same time as running the ping to the new server I had also left ping running to another of the old servers and there is no loss of connectivity...

I have run diagnostics on the dell server NIC and eveything checks out ok. I have changed the switches back to the old ones, just in case there was a problem with the new ones. I have been only using NIC 1 on the server and tried changing to NIC 2, same thing. I am now running out of ideas!!


Any suggestions/advice would be greatly appreciated.

 

[netmanrick]

What about the load on the new server: memory;disk io;tasking;etc.
If the new switches are managed,use the sw to monitor the server port.
What about nic/switch port setup? Did you hard code the speed/duplex or accept auto?

Rick Harris
SC Dept of Motor Vehicles
Network Operations

 

[bcastner]

Set a higher Time To Live for the command and you likely will see 100% responses.

ping -i TTTL host

 

[DylanThompson]

I would also be checking the event viewer of the new server to see if there are any events that correspond with the timing of the delayed response from the ping. If you can find some correlation between an event or events on the new server and the delayed response you may be able to get a better idea of what is going on.

For example, I have seen AD corruption cause a very slow period of network activity every five minutes in a mixed 2000 and 2003 domain.

One other thing that seems to make a lot of difference. If you can set the duplex rate for the port on the switch that server connects to, change to 100/Full (or 1000MB/Full if you have GB LAN) and also change the settings for the NIC on the server to match this duplex rate. I have seen this change make drastic improvements to network access speeds in general.

 

[DTSMAN]

I am having the same problem with a Dell that is running xppro that functions as a server on a POS system. I have checked everything also and it is becomming an issue because the software that I run throws up and error message everytime it happens. Sometimes it runs for 10 minutes and sometimes for 2 hours. hmmmm..

We have changed out the hub and the cabling, everything points to the Dell.

Bo

I'm a man,, I can change,, If I have too,, I guess. (Red Green)

http://www.redgreen.com/

 

[RatherBeSkiing]

Thanks for your replies, there is very little load on the server, currently its not being used to its potential because of this issue.

There is nothing in the event log that coincides with the loss of network connectivity.

I have tried changing patch cables, sockets, etc... The server is now plugged directly into the switch. I also tried specifying the speed of the NIC instead of Auto and interestingly when set to auto the switch display shows connection at 100Mbit Full duplex. When the NIC is set specifically to 100Mbit Full duplex the switch only shows connection to be 100Mbit half duplex?!

I'm starting to think the same as Bo, things are pointing to the dell kit

 

[bcastner]

Leave the setting at AUTO. Negotiation issues between NIC and switch are common with fixed speed and duplex settings at the NIC unless you can force all connections to manual settings and make them consistent.

For 100mbs networks, consider some registry settings described in this MS KB:

http://support.microsoft.com/?kbid=315237