1820622 Members
1869 Online
109626 Solutions
New Discussion юеВ

package location

 
JBro
New Member

package location

right now I have a simple package that runs an application. The application is in:
$SGCONF/mypackage/
it could be an exe, or a script. It writes data to the LVM/VG (raid).
The package will eventually have more exe/scripts - I am migrating stuff done in:
/opt/theapp
to the package.
I would like to move everything to to an 'app' LVM. That way, I do not have to worry about maintaining the app over several servers.
So my question is....what is the acceptable package guts location?
1) $SGCONF/package
2) /opt/theapp [leave it there]
3) (LVM)/app

thanks!
-john
2 REPLIES 2
Thomas J. Harrold
Trusted Contributor

Re: package location

John,

There are many ideas on the "best" location to keep your applications, but I prefer to keep a separate copy on each node in your cluster, for redundancy.

I use rdist or rsync utilities to make sure that they are sync'd up, so I don't have any surprises when the package fails to a backup node.

For the SG configuration and control scripts, keep them in their standard location, typically /etc/cmcluster.
Again, I use rdist/rsync to keep the control scripts syncronized between cluster nodes.

-tjh
I learn something new everyday. (usually because I break something new everyday)
A. Clay Stephenson
Acclaimed Contributor

Re: package location

The answer is that it depends upon the package but, in general, I tend to put the application executables and the data in a shared VG and let all of them move with the package. The advantage of this approach is that you patch the application in one place and its patched for all nodes. There is a "gotcha" and that is shared libraries. If for example, a new shared library that is dynamically linked to your application is installed on the active node, you must make certain that this same shared library be installed on all the potential adoptive nodes or there is no assurance that the application will even start (or behave the same) on the adoptive nodes. In any event, it always makes sense to keep all the nodes in a cluster at very nearly the same patch levels.
If it ain't broke, I can fix that.