I'm still confused with a scenario on our client infrastructure. They have two serveur bd1 and bd2 connected directly on a IBM V3700 Storage.
They want to create a file system on bd1 and share this one with bd2 so when they will create a file on the mount point of the shared file system on db1 they can access it on db2...
It's my first time to work on AIX.... I have no documentation to guide me doing what our client needs.
If someone can explain to me a way or give me a step by step link where i can do it easily i will be very happy.
I'm not an AIX expert so I was hoping that an AIX specialist but I can offer some generic principles on this.
Firstly, you can only ever mount any one filesystem on one server. Only one server can control the superblock (free space table), file locking, and file writing. To have more than one server thinking they could do that is a recipe for instant filesystem corruption.
So there are two common ways to generate the scenario you outline.
Server BD1 can mount the filesystem (like a local drive even though it's a LUN from a SAN), then publish a Network File System (NFS) handle allowing BD2 to mount the same filesystem remotely. Therefore, all I/O to/from the filesystem is going through BD1 which has control. The problem with this "poor man's" solution is that there is no resilience if BD1 goes down.
A "high end" solution would employ a clustering suite. There are many of these available offering different features. In such a configuration each server usually has its own hard IP address and also a virtual IP address that the clients call in on. However, generically speaking, the filesystem is physically connected to both BD1 and BD2 at the same time (via FC, iSCSI, or whatever protocol) with the cluster suite arbitrating which one has control and handles all the I/O. Between BD1 and BD2 there are heartbeats arranged (could be via a serial connection, over IP, or via a file on the SAN) so that one server can detect if the other goes down. In the event of a server failure, the survivor will mount the orphaned filesystems, check them for consistency, switch over the dead server's virtual IP address to itself, and start the applications that the lost server was running. The clients can then continue working within a few short seconds.
Don't be confused by the IBM V3700's capability to remotely mirror a filesystem independently of any host. The V3700 is an entry level enterprise storage device with extreme versatility, BUT, it remains that any filesystem can only be under the control of one host at a time. In some big end systems the filesystems can be controlled (mount, file locking, writing, caching) by the storage controller itself but you are talking very big money.
Hope that helps and that you will get a better answer from an AIX specialist on here.
