Switch upgrades and red NICs

[crispybits]

Hi everyone

 

My network manager is away on holiday right now, and I'm having a very weird (to me) problem.

 

We've just upgraded our network to 10G fibre and installed a load of new HP 2920 switches. The default vlan is set up on them but not much other config has been done.

 

Several computers around the school (more and more with each passing day) that were previously able to log in via the new switches are now getting no connection to our login server, or at most an intermittent connection (they will log in fine then 2 hours later no login server is present). The NIC light on the back has also changed from green to red in the majority of cases.

 

I've tried googling but the solutions I have found are all related to home networks and unmanaged switches. Seeing as the issue is affecting multiple PCs all over the site I don't think those answers will be particularly relevant (though I am prepared to try one or two with reinstalled drivers if that's worth it?). Is there something in the out-of-the-box switch config that may be causing this?

 

Any hints or tips in how to investigate an resolve this issue would be very much appreciated!

[IanT]

Hi,

 

Whats the error you are seeing on the workstations?

 

Can you log into the switches via the CLI and do a show conf?

 

Are you getting a DHCP Address on the workstations?

 

Can you ping the core switch?

[crispybits]

I can ping the core from the switches. Ping times coming in around 3-4ms. Pinging from a workstation directly plugged into the core gives me a 1ms ping to the switches.

 

I am getting a DHCP Server IP address from ipconfig /all on the workstation

 

The error when I try to login one of the PCs that's gone down is the standard "There are no logon sevrers available to process your request" as if the ethernet cable is unplugged (I have checked they are still plugged in locally and patched at the switch before I started the thread)

 

Here is the config from the show conf on one of the new switches:

 

Startup configuration: 5

 

; J9728A Configuration Editor; Created on release #WB.15.17.0007

; Ver #07:eb.8f.fc.f3.ff.37.2d:89

 

hostname "A5-2920-48G"

module 1 type j9728a

timesync sntp

sntp unicast

sntp server priority 1 XXX.XX.XX.20

ip default-gateway XXX.XX.XX.254

ip ssh filetransfer

snmp-server community "public" unrestricted

snmp-server community "mw-snmp" operator unrestricted

snmp-server community "mw-snmp-read" operator

snmp-server host XXX.XX.XX.6 community "public"

snmp-server host XXX.XX.XX.20 community "public"

snmp-server host XXX.XX.XX.204 community "mw-snmp" trap-level all

snmp-server host XXX.XX.XX.14 community "mw-snmp-read"

snmp-server host XXX.XX.XX.219 community "mw-snmp-read"

snmp-server contact "MSh" location "C1-Server Room"

oobm

ip address dhcp-bootp

exit

vlan 1

name "DEFAULT_VLAN"

untagged 1-48,A1-A2,B1-B2

ip address XXX.XX.XX.73 255.255.252.0

exit

no tftp client

no tftp server

no autorun

no dhcp config-file-update

no dhcp image-file-update

password manager

 

(apologies for the XXXed out IPs, I'm not a network guy so I don't want to accidentally publish something on the internet that may cause a security issue - the numbers under the XXXes are what I would expect to see)

[PotNoodleTech]

Sounds to me like you're PCs can no longer connect to the DHCP server, so when their lease runs out - they drop off the network as they can't get an IP address.

 

Were you're servers on separate subnets or VLANS or was there any routing configured on the old switches?

 

Ps. the red light on the back of the nics could just be the nic showing it is now connected at 1000mb instead of 100mb?

Edited August 7, 2015 by PotNoodleTech

[crispybits]

Yep there were VLANs set up previously on the old switches - we had a switch config guy come in the other day and he removed a load of VLANs from our network he said he thought were legacy and not used any more - does this mean it's likely to be the servers were on one of those VLANs and now we'll have to do some config on them? He's back in on Tuesday and obviously it's the holidays right now so if it's something like that I'll leave it until he comes back in. I'm a PC tech so while I know the basics of networking I'm not knowledgable enough to tackle a problem like that on a network the size of the one we use unless it's simple enough for someone to give me a step by step in here.

[PotNoodleTech]

Sounds like trouble. Not really possible to give a step by step when you don't know what your config was. There's no documentation or anything anywhere that might shed any light on how things were set up? Do you have the old switches that you could connect to and look at the config?

[crispybits]

I've got details of all the old vlans - he saved them into a text file before he started removing them. Not got full config from the old switches tho. We've reassigned most of the fixed IPs from the old switches directly to the new switches so I think it would be very likely to cause an IP conflict if I powered one back up again to check things.

 

If it's likely to be something along these lines I'll just wait for the specialist to come in on Tuesday - the one thing that bugs me more than anything is unqualified/amateur people causing more damage by trying to fix problems they don't fully understand and I don't want to drop into that category myself. The staff have a remote login that gives them 90% functionality from off site which appears to still be working OK so no huge rush from that side of things.

Edited August 7, 2015 by crispybits

[IanT]

As you say I would wait for the network guy to come back in.

 

Just looking at the config you pulled off there seems to be a lot of work that needs doing as that's a very basic switch config

[crispybits]

OK thanks for the help - at least I know I can stop googling every variation of the problem trying to work it out now