[Linux-cluster] Need help for Clustered NFS

[Lon Hohberger]

On Wed, 2005-07-20 at 10:10 -0400, haydar Ali wrote:

> mkfs -t ext2 -b 4096 /dev/sdd2
> mkfs -t ext2 -b 4096 /dev/sdd3

The quorum partitions shouldn't have a file system.  Cluster manager
will destroy all the data on those file systems.


> I created another huge partition /dev/sdd4 (over 600GB) and formatted it in 
> ext3 filesystem.
> 
> I installed the cluster suite on the 1st node (RAC1) and I started the 
> rawdevices on the two nodes RAC1 and RAC2 (it's OK).

Looks like your configuration is from clumanager on RHEL 2.1 AS (there
was no RHCS for RHEL 2.1).


> This the hosts file /etc/host on the node1 (RAC1)
> 
> # Do not remove the following line, or various programs
> # that require network functionality will fail.
> 127.0.0.1              localhost.localdomain localhost
> #
> # Private hostnames
> #
> 192.168.253.3           rac1.domain.net     rac1
> 192.168.253.4           rac2.domain.net     rac2
> 192.168.253.10          rac1

> [cluster]
>   alias_ip = 192.168.253.10

The cluster alias IP is only for monitoring, not for services.  In any
case, you need to have different hostnames for the cluster members, the
alias IP, and any service IPs you intend to use.


> [powercontrollers]
> start powercontroller0
>   IPaddr = rac1
>   login = unused
>   passwd = unused
>   type = null
> end powercontroller0
> start powercontroller1
>   IPaddr = rac2
>   login = unused
>   passwd = unused
>   type = null
> end powercontroller1

Obligatory note: Please buy a power controller.


> [services]
> start service0
>   checkInterval = 30
> start device0
> start mount
> start NFSexports
> start directory0
> start client0
>   name = rac1
>   options = rw
> end client0
>   name = /u04
> end directory0
> end NFSexports
>   forceUnmount = yes
>   fstype = ext3
>   name = /u04
>   options = rw,nosuid,sync
> end mount
>   name = /dev/sdd4
>   sharename = None
> end device0
>   name = nfs_project
>   preferredNode = rac2
>   relocateOnPreferredNodeBoot = yes
> end service0

> And when I launched this command on RAC2:
> mount t nfs rac1:/u04 /u04
> It list the following error message :
> Mount: rac1:/u04 failed, reason given by server: Permission denied
> 
> Can someone help me to fix this problem in this configuration?

a. Your /etc/hosts needs to be fixed: rac1 will first resolve to
192.168.253.3, so that server would refuse the mount.  rac1 needs to be
one IP, and the .10 address needs to have its own hostname.

b. Your service needs its own IP address.  I'd scrap the cluster alias
IP and move it in to the service.

c. You don't have RAC2 as a listed client in the service.  So, if (a)
wasn't true, mounting from RAC2 will always fail because it's not in the
NFS ACL.

Good luck!

-- Lon