Cluster nodes cant ping each other

Question

Hi, I have recently put one of our old clusters back online, its a poweredge 1855. I'm working on 5 nodes, and it has a PowerConnect 5316M on the back. The 5316M has only 1 cable connecting it to the uplink switch. Each blade has 2 NIC's. One of them is a normal production NIC, with a normal IP on our range. All blades can ping each other on this address, and they all have internet. The other NIC is what I'm going to use for live migration, or cluster heartbeat or something ..either way, it has a different range with no gateway (172.50.1.x) Now, 3 of the blades can all ping eachother, but not the other 2. And the 2 that cant ping, can ping eachother. So it looks like 2 are on the same VLAN, and the other 3 are on the same VLAN. The problem i have, is the switch shows them all on the default VLAN ..and one thing i cant work out ..is if the servers all have 2 NICs, but there are only 10 internal switch ports? ..so how can i set either NIC on a different VLAN? The switch config is here: console# show runspanning-tree mode mstpspanning-tree mst configurationinstance 1 add vlan 1,15,900instance 2 add vlan 2-3name PDSNetherlands1revision 900exitspanning-tree mst 1 priority 57344spanning-tree mst 2 priority 57344interface port-channel 1description 'LACPGROUP-M5316M-SW2'exitinterface port-channel 1switchport mode generalexitvlan databasevlan 2-3,15,238,328-329,501,638,900exitinterface port-channel 1switchport general allowed vlan add 15exitinterface port-channel 1switchport general allowed vlan add 900exitinterface vlan 2name 'PDS Internet'exitinterface vlan 3name 'PDS Wireless'exitinterface vlan 15name 'PDS ServerInfrastructure'exitinterface vlan 238name 'SHELL GI-Desktops'exitinterface vlan 328name 'SHELL Unix-Desktops'exitinterface vlan 329name 'SHELL Windows-Servers'exitinterface vlan 501name 'SHELL Linux-Desktops'exitinterface vlan 638name 'SHELL IP-Telephony'exitinterface vlan 900name 'PDS Workstations'exitinterface range ethernet g(11-16)channel-group 1 mode autoexitinterface vlan 1ip address 10.254.254.253 255.0.0.0exitip default-gateway 10.0.0.1logging 10.0.1.56 severity debugging facility local0username admin password 5ebe2294ecd0e0f08eab7690d2a6ee69 level 15 encryptedusername admin_jon password ea4462b55746d29a4a0fc44c3db06b95 level 15 encryptedusername administrator password 22d4734b069e91ceebdf297701005b28 level 15 encryptedsnmp-server community private rwsntp client poll timer 60sntp unicast client enablesntp anycast client enablesntp broadcast client enablesntp server 64.90.182.55 pollconsole# Any help would be appreciated, I cannot complete the cluster validation wizard until i can crack this .. Kind Regards, Jon

JonPDS · Answer

Hi Daniel,

Thanks for the reply. Here is more information:

We have 2 x 5316 installed in the back of the chassis. Each of them now has a single cable in external port 14 connecting it to our core switches, these are just on VLAN 1.

The servers that I'm currently working on are installed in slots 5,6,7,8 and 9 in the front of the chassis.

Slots 1,2,3 also contain servers, but I'm not too concerned about those right now. Slot 4 is just empty.

When I run the above command, the result is the same for all internal ports ...VLAN 1, Access mode, Gvrp Status: disabled ...no forbidden VLAN's.

I just cannot see any different in the ports, but 6 and 7 can still only see eachother.

One thing i'm confused about ...if there are 2 modular switches in the back, with a total of 12 external ports (to connect to various external networks) ..and they have a total of 20 internal ports ...how do i know which interface on which server is connected to which internal port?

All of the blades have 2 NIC's, but i dont have a way to tell which internal port they go into?

Thanks so far!

cheers,

Jon

JonPDS · Answer

Hi Daniel,

Thanks for that link, I had a read through it. I can understand what you're saying about NIC 1 going to module 1 and NIC 2 going to module 2 ..that makes sense now.

All the servers are using the same NICs for the same type off network ...NIC 1 on all blades is the 10.x.x.x production NICs ..all of them can see the rest of the network. Then the secondary card is the 172.x.x.x range ...but this is only working for 3 nodes. ...all the 172.x.x.x addresses are on port 2.

The switches look largely unconfigured ...that is all the ports are on VLAN 1.

So does this mean for blade 1, NIC 1 goes to port 1 on I/O module 1 ...and blade 1 NIC 2 goes to port 1 on I/O module 2?

Sorry for the confusion but I'm not sure why two of the 172's cant ping the rest?

Cheers,

Jon

JonPDS · Answer

Hi,

this is fixed now. Indeed it was the internal ports on the Switch-2 which were configured for a different VLAN. The CLI didnt show this, and the normal VLAN membership GUI wouldnt let me change them ...the VLAN port settings would let me change the VLAN tho. .

all are talking now ..thanks for your help

Regards,

Jon

Networking General

Was this post helpful?