making nfs package.

[khilari]

Hi people, okay so I am configuring an nfs package. The cluster is up and running. I am attaching the package files out put. Kindly tell me whats the next step to do.


#Vi nfs.conf (configuration file)

PACKAGE_NAME nfs.conf

PACKAGE_TYPE FAILOVER

FAILOVER_POLICY node2

FAILBACK_POLICY MANUAL

NODE_NAME node1

NODE_NAME node2


RUN_SCRIPT /etc/cmcluster/nfs/nfs.cntl
RUN_SCRIPT_TIMEOUT NO_TIMEOUT
HALT_SCRIPT /etc/cmcluster/nfs/nfs.cntl
HALT_SCRIPT_TIMEOUT NO_TIMEOUT


AUTO_RUN YES

LOCAL_LAN_FAILOVER_ALLOWED YES

NODE_FAIL_FAST_ENABLED YES

SERVICE_NAME nfs1.monitor
SERVICE_FAIL_FAST_ENABLED NO
SERVICE_HALT_TIMEOUT 300


SUBNET 142.221.210.0


Control file:

# cat nfs.cntl
PATH=/sbin:/usr/bin:/usr/sbin:/etc:/bin
VGCHANGE="vgchange -a e" # Default
VG[0]=vg01 LV[0]=/dev/vg01/lvol1;
FS[0]=/hanfs/vg01; FS_MOUNT_OPT[0]="-o rw"
FS_UMOUNT_COUNT=1FS_MOUNT_RETRY_COUNT=0
IP[0]=142.221.210.43
SUBNET[0]=142.221.210.0

# cat hanfs.sh
XFS[0]=/hanfs/vg01
NFS_SERVICE_NAME[0]="nfs1.monitor"
NFS_SERVICE_CMD[0]="/etc/cmcluster/nfspkg/nfs.mon"

NFS_FILE_LOCK_MIGRATION=0
NFS_FLM_SCRIPT="${0%/*}/nfs.flm"

The directory structure is like this:

# ll
total 48
-rw------- 1 root root 5676 Apr 7 14:10 cmclconfig
-rw------- 1 root root 0 Apr 10 12:37 cmclconfig.tmp
-rw-rw-rw- 1 root sys 27 Apr 7 13:48 cmclnodelist
-r-------- 1 bin bin 0 Mar 15 2004 cmknowncmds
---------- 1 root root 0 Apr 7 14:04 config.lck
dr-xr-xr-x 2 bin bin 96 Oct 26 03:42 cvm
drwxrwxrwx 2 root sys 8192 Apr 10 13:36 nfspkg
drwxr-xr-x 2 bin bin 96 Oct 26 03:30 sharedtape

And the nfs configuration, control and hanfs.sh files are under nfspkg directory.

Kindly, check to see if I have to add any thing for this nfs package. My concerns are with the package name, can it be anything or it should be nfs.conf, because I did get errors when I was cmapplyconf for the package.


If u have can u kindly, send any other applications toolkit or both configuration and control scripts. Like that for a clock, or any other inbuilt processes for practice.

Thanks.

[melvyn burnard]

1) Use:
cmapplyconf -P /etc/cmcluster/nfspkg/nfs.conf

2) Read the README for the NFS toolkit in /opt/cmcluster/toolkit/nfs

3) Read the Managing Serviceguard manual

4) Read the NFS toolkit manual for your version available at:
http://docs.hp.com/en/ha#Highly%20Available%20NFS

[Steve Lewis]

I would also add "async" to the mount options.

[khilari]

i dont have an nfs toolkit, can i make it without a toolkit and can u tell me of any other type of package that i can make without having a toolkit.

[A. Clay Stephenson]

You can make any kind of package completely on your own. Some packages (like NFS and databases) have already been done so that the Toolkit will save you time and are likely to be more robust than your initial attempt. For many (most) applications, no toolkit exists so that you do those totally on your own.

MC/SG is really something that is best learned in a class; not so much for MC/SG itself but rather to bang into your head that MC/SG is the last thing you do after making your systems so robust that MC/SG itself is hardly needed.

[Kent Ostby]

What errors did you get when you did the cmapplyconf?

[Stephen Doud]

I don't have any sample scripts from which to create a test package.
However, you can read the package control template (cmmakepkg -s
) and see that if your server has xclock, you can start an xclock service, and send the display to a remote system that can display X-based images.

Otherwise, don't put anything in the package control script, and simply configure it's partner package configuration file to point to it.

This will allow you to create a sample package that you can start and stop (although it won't do anything useful beyond that).