[c-nsp] Windows server hangs connected to 3750

Vigar, Damien Damien.Vigar at det.nsw.edu.au
Wed Nov 26 20:26:14 EST 2008


Responses inline...

>
> Damien,
>
> What sort of troubleshooting have you done thus far ?
>
> Some questions/suggestions:
>
> * check to make sure MAC addresses on the servers aren't the same. Sounds
> strange but I've had it before (from a reputable vendor whose name is
> abbreviated to 3 letters).

No, they're unique - I've heard of this but never seen it. The NICs show as HP NC373i Multifunction Gigabit Server Adapter on our newer server, and HP NC7781 Gigabit Server Adapter on the server at another smaller site where the issue has occurred.

> * hard code both switch & server to 1000/full to make sure you're not
> getting speed/duplex mismatches.
As mentioned in another email, we've tried both auto and hard-coded. Currently the local support guys onsite have moved the server back to a fast ethernet port, and the problem hasn't recurred (ports and NIC both on Auto-negotiate)

> * does the problem continue to happen when ONLY the servers are connected to
> the switch (ie. isolated from rest of the network) ?
I'd love to be able to test that - as these are the DC/file & print servers for TAFE (educational) campuses we won't be able to do that, at least during the week. And as a government job, overtime doesn't happen :-)
Hmm - we've got the Xmas break coming up soon, I may be able to get them isolated for a bit during this time...

> * try new GBICs to check they aren't bad. Swap the ones that aren't working
> for some that are known to work (you mentioned you have some
> switches/servers that are working).
I can try this. It'll take time - we have one server at each of our 35 sites, spread across about half of New South Wales. So I can send new GBICs to sites and get some returned, but it might not be for a while.

> * change the NIC in the server. You're probably using the onboard NIC's so
> this would mean adding a PCI Gb NIC to a server to test.
Pretty sure the onboard card is what is in use. I'll see if there's any NICs around that we can test with.

> * move one of your servers that IS working to the same switchport that isn't
> working with these new servers. This will pretty much rule out the switch
> (provided that it continues to work) and you can start looking at the
> servers harder.
As mentioned above, we've pretty much only the one server at each site. I'll check what we can do, though.

> * does it happen immediately (ie. as soon as you connect the servers) or
> does it takes minutes/hours for the problem to show up ?
It can take quite a while. It happened about once a week for a couple of weeks.

> * have you got anything fancy configured on the switch ?
Pretty basic - running one vlan for data, another for voice (Cisco IP Tel).
The SFP port the GBIC is in is configured only for the data VLAN.

> * To attempt to see if it's a software issue you could boot up a live linux
> distro (eg. knoppix) and see if it has the same problems with the Gb
> cards/switch. If that works fine, then a reinstall of Windows might be
> called for and following that up with MS tech support if problem still
> occurs.
I'll speak to our local support staff and see if we can get a box booted with knoppix or similar put onto the port for a while.

> * I think someone else already asked if you're doing anything fancy like
> clustering or load sharing/balancing ?
Yep. Nothing like that at these sites.

> regards,
> Tony.

Cheers,

Damien

**********************************************************************
This message is intended for the addressee named and may contain
privileged information or confidential information or both. If you
are not the intended recipient please delete it and notify the sender.
**********************************************************************


More information about the cisco-nsp mailing list