AIX

Hi,

My NFS does not work till I start HACMP. The NFS service is indeed started when server starts. But it is not in active state, when I use lssrc to check NFS services' status I find they are all in inoperative status. But after I start HACMP (smitty clstart), all NFS services become active automaticlly.

Why is that ?

BTW, my /etc/exports is rather simple, something like

/share -anon=500,sec=sys:krb5p:krb5i:krb5：dh,rw

Because of the HACMP design. HACMP does not expect you to have exports beside the cluster resources. Using both the AIX /etc/exports and the Cluster /usr/es/sbin/cluster/etc/exports is a (well known) possible source for cluster failure i.e. bad administration. Without the activated Resource Group there should be nothing of interest on the cluster node. So do not use a non-service IP-address or a persistent address for NFS exports. You should use a Service Label for the NFS export and let HACMP handle when/where to export the FS.

Hi, thanks for answering my question!

I am a totally newbie to HACMP, so can you further elaborate how I can use Service Label for the NFS export ?

Thanks again!

Well, you were a newbie in last October (Question about HACMP for active-active mode) but now is almost March. Fortunately HACMP/PowerHA is pretty well documented. Have a look at the Installation Guide at the IBM Cluster information center and then at the HACMP Cookbook. IBM Redbooks | PowerHA for AIX Cookbook. Good luck

Oh hey, thanks!

I paste the procedures here in case someone else also needs it. So in the cookbook chapter 2.4.3 , it mentions

1. Shared volume groups that contain file systems that will be exported by NFS must have the same major number on all nodes, or the client applications will not recover on a fallover.

2. If NFS exports are defined on the node through PowerHA, all NFS exports must be controlled by PowerHA. AIX and PowerHA NFS exports cannot be mixed.

3. If a resource group has NFS exports defined, the field “file systems mounted before IP configured” must be set to true.

4. By default, a resource group that contains NFS exported file systems, will automatically be cross-mounted. This also implies that each node in the resource group will act as an NFS client, so must have a IP label on the same subnet as the service IP label for the NFS server.

BTW, the reason I still a newbie is that this is NOT my job responsibility at all. We have some specific group people who are supposed to do this for us, but they are doing a lousy job.