Does anybody know of any reason why this might be happening. The
current configuration is 1 Ethernet port on each VIOs. I have 2
virtual ethernet adapters on each VIO, the 1 we were using before we
started testing Q-tagging and the new one with the VLANs. I used
chdev to add the new V-ethernet adapter to the existing SEA. All four
virtual adapters are configured as trunk adapters and they have a
trunk priorities ranging from 1-4.
Currently I have no client LPARs even turned on in an attempt to
narrow down the problem. Any help on this is greatly appreciated.
Hi
I'm pretty sure that you have forgotten the control channel for the
SEA.
You must have a control channel when you try to setup
SharedEthernedAdapters with failover inbetween the VIOs.
Currently your setup is trying to create a spanning tree loop bpdus'
comming in via one of the VIOS and out the other, when the switch
notice this it will disable the port. (Be happy that the network guys
configured the port this way, i have tried it on a big network once
without this guard)
Do like this.
Basic setup with SEA failover inbetween two Virtual I/O servers
assign three virtual adapters to each VIO server.
adapter 1 plain adapter no trunking no nothing, put it on your
administrative VLAN (let's say 100)
adapter 2 trunk adapter on VLAN 1 802.1q add your VLANS
(100,101,102,130 for example) on primary virtual I/O server Priority 1
secondary Priority 2
adapter 3 ctl_channel adapter on VLAN (any VLAN that your not trunking
out) for example 99
lsdev |grep ent would look something like
ent0 Available 2-Port 10/100/1000 Base-TX PCI-X Adapter
(14108902)
ent1 Available 2-Port 10/100/1000 Base-TX PCI-X Adapter
(14108902)
ent2 Available Virtual I/O Ethernet Adapter (l-lan) <-
plain adapter for the VIO server
ent3 Available Virtual I/O Ethernet Adapter (l-lan) <-
Trunk Adapter
ent4 Available Virtual I/O Ethernet Adapter (l-lan) <-
Ctl_channel
then just create the SEA on your primary VIO server
mkvdev -sea ent0 -vadapter ent3 -default ent3 -defaultid 100 -attr
ha_mode=auto ctl_chan=ent4
Run the same command on the secondary VIO server
Then test your setup if it's redundand by issuing the command:
chdev -dev ent5 (SEA interface) -attr ha_mode=standby
on your primary VIO server
Check your errorlog (errlog|head -10)
40D97644 0905043208 I H entX BECOME BACKUP
Your secondary will become primary in the errorlog
configure an IP on ent2 in above example and ping the gateway to
verify that it's working as it should.
One other tip. If your on VIO 1.5.0.0 upgrade to atleast fp 10.1
before you do anything.
If it doesn't work on default VLAN 100 it's very likely that the
network crew have configured a native VLAN on the port/etherchannel
and VLAN 100 is untagged when it gets to the hypervisor switch.
Good luck
Jesper