This content has been marked as final. Show 4 replies
A couple of things:
- OSC's logic with the public network relies on IPMP, even if there is only one network "interface" behind it. So removing IPMP won't help
- You are saying that the link aggregation in the IO domain works well?
- Are you saying that the vnet within the guest works well as well? But IPMP complains about the interface being down?
- Can you use the underlying interface with no problems from within the guest?
- Did you try configuring IPMP in the IO domain instead of a link aggregation? Does this help?
What are the IPMP error messages in /var/adm/messages?
I do not think this is a cluster issue, but a networking one - either with LDOMs or Solaris. OSC is only interested in the status of the IPMP group, but does not set it. That is done by the base OS (in.mpathd)
Thanks for the reply.
Yeah, it was a network config issue caused by SC because it changed /etc/hostname.vnet2 to -failover during cluster configuration. So after replacing failover with up, the IPMP group is showing up now. This is weird behaviour because I have another 2 nodes cluster in guest ldom setup the same way with this same configuration for public (one interface) IPMP but that one did not cause any issue.
Thanks for the help though.
nfsp0gux001 group sc_ipmp0 -failover
nfsp0gux001 group sc_ipmp0 up
IPMP issue with one interface in guest LDOM.
So it seems this is not easy to reproduce? If you ever happen to see this again during the OSC installation, please remember your steps and let us know. This must not happen as part of the installation process.