baseline switch, sql disconnects and port up/down

[netspeed]

I have two customer sites with baseline switches. One is a 2928-SFP Plus with latest firmware.



The problem is, both sites have a sql based program running (different problems) and they lose connection to the sql database intermittently, a different user each time.



The logs on the switch say things like:

Nov 12 09:57:37:667 2009 IFNET Warning LINK UPDOWN GigabitEthernet1/0/11: link status is UP

Nov 12 09:57:35:204 2009 IFNET Warning LINK UPDOWN GigabitEthernet1/0/11: link status is DOWN



This happens intermittently on random ports. There is not a loop issue and if they are moved to another non 3com switch everything works fine.



The link only goes down for a second so not enough time to register in windows that its down and everything else on the network works fine but the sql database seems to be less forgiving and its enough time to kick them out and they have to kill the software and start it up again.



The switches have no config on them, they are out of the box with network cables plugged in. No settings are changed. Has anyone ever come across this before?

[Fred_Mancen_1]

Try to set up the ports hard coded in 1000/full-duplex; and configuring the NICs on the server in the same way.



HTH

[netspeed]

Well, lets assume that if I did that I find the problem goes away. It is not something I should have to do and not all computers are going to be 1gb. Also tracing each computer to a particular port on the switch is time consuming, impractical and again there should be no need for this from a switch.

[Fred_Mancen_1]

OK, I agree with you that this is not a behavior we should expect from a switch...but, believe me, many times I faced the same issue and the workaround was always the same. Even with switches from other vendors.

A couple years ago, a Sun Fire 15K server (a lot of money, right?) was not connecting at 1G/full at a Cisco Catalyst 6513 switch, only in half-duplex. After two days, my team figured out that was a Cisco known issue concerning that server NIC chipset when working with that type of module. Only that connection was forced to be moved to another switch.



These things happen...



Regards