HP Procurve 2610 48 PWR PoE - VLAN Drop outs

[MDKB]

Hi

We recently setup VLAN 10 and VLAN 20 on a HP Procurve and am getting some odd behaviour that no one can solve. Basically VLAN 20 network (10.10.20.0/24) can reach VLAN 10 (10.10.10.0/24) fine but going between VLAN 10 and VLAN 20 there are long periods of drop out. From the switch itself I can ping the VLAN 20 gateway when it happens but cannot reach anything beyond it. These drop outs occur for periods of about 5-15 minutes and then it works for 5-15 minutes and then drops out again. Its been a nightmare to resolve. A Copy of the config is below. Any help would be appreciated. I am starting to think it maybe the switch at fault. We have third party Juniper firewall support (10.10.10.254) which is the gateway to the internet but they have put everything in place that should enable the VLAN traffic to go both ways and they say its setup correctly.

We have two other HP switches but I moved relevant machines to the main switch in order to isolate the issue as can be seen in the config.

hostname "HP-2610-1" 
max-vlans 24 
qos dscp-map 101110 priority 5 
interface 1 
   name "uplink to TeleTellus Pro (10.10.20.9)" 
exit
interface 49 
   name "uplink to Phone Server (NEC)" 
exit
interface 50 
   name "uplink to 10.10.10.109 (HP)" 
exit
ip routing 
snmp-server community "public" Unrestricted 
vlan 1 
   name "DEFAULT_VLAN" 
   untagged 50-52 
   no ip address 
   no untagged 1-49 
   exit 
vlan 10 
   name "DATA_VLAN" 
   untagged 2-48 
   ip address 10.10.10.102 255.255.255.0 
   ip helper-address 10.10.10.210 
   tagged 50 
   exit 
vlan 20 
   name "VOICE_VLAN" 
   untagged 1,49 
   ip address 10.10.20.254 255.255.255.0 
   qos dscp 101110 
   ip helper-address 10.10.20.200 
   tagged 2-48,50 
   voice 
   exit 
qos type-of-service diff-services 
ip route 0.0.0.0 0.0.0.0 10.10.10.254 
spanning-tree
password manager

[Arimo]

Hi

You have separate voice and data VLANs. Your complete telephony system seems to be connected to your voice VLAN. Since the idea of doing this is to separate voice and data traffic, why do you want to route traffic between these VLANs in the first place?

[MDKB]

so that the PC with telephone reporting software can communicate with the NEC server.

and so the users on 10.10.10.0 network can access their voicemail admin features on the NEC server.

basically the phone server on 10.10.20.0/24 needs to be able to be reached by various PC's on 10.10.10.0/24

I have found a work around by putting  a second NIC in the server with 10.10.20.0 network ip address. This at least gives me a way onto that network when I need it.

I am going to speak to HP support now as I believe this issue is a fault on the switch since various tests point to it. maybe arp or something.

[Oleg Sukharev]

Hi, MDKB

I guess ports 2-48 should be _untagged_ in vlan 10 and tagged in vlan 20, so PCs connected to that ports can access VLAN 10 easily.

Is this switch is the main router for VLANs 10 and 20?
Who is the default gateway for hosts in VLAN 10?

Also you can omit adding ip helper-addresses in VLANs pointing to DHCP from the same VLAN.

hth,
oleg