2009-09-16 08:37 AM
Stephane, I found the following information for you on our NOW site:
You can add a disk shelf to an active/active configuration without powering down your active/active configuration (sometimes referred to as hot-adding the disk shelf), as long as your system meets the active/active requirements.
Hot-adding disk shelves enables you to add (but not swap) disk shelves without a service interruption. However, you cannot add more disk shelves to your active/active configuration than it can support, either for the entire configuration or for either node. See the System Configuration Guide at the NOW site for maximum storage capacity values.
Is that the information you were looking for?
2009-09-16 08:41 AM
There are going to be two parts to this answer. First let me ask you; do you already have additional shelves on the FAS270?
If the answer is yes, then this can be preformed as a non-disruptive upgrade. Assuming that the newly introduced shelves meet all of the necessary conditions (no existing ownership being the primary concern) and your loop is not already at its limits.
If the answer is no, then you are going to need to halt the head (not the cluster if it is a FAS270C), and change the termination switch on the head. In the case that there are no additional shelves attached to the head(s), the termination switch will be set to on. The FAS270 is not auto terminating, thus the need for the switch. In the earlier models, it is more of a "dip switch" that you will need to use some type of tool to get at. In later models it is a surface mounted switch that needs no tools.
Keep in mind that as long as you have a cluster license and cluster fail over is enabled, you can still maintain data availability through the partner.
Best of luck to you.
2009-09-16 11:57 PM
Actually, it's a FAS270C and I don't already have any additional disk shelf.
But when one of the heads fail, even if data are OK, user connections are broken during the "takover" operation, isn't it ?
2009-09-17 09:05 AM
It depends on the protocol that you are using. If you are using FCP or iSCSI, and you have multiple paths enabled (with MPIO or equivelent) then it will fail over and not have a service interuption. If you are connecting via CIFS, NFS, HTTP, FTP, or otherwise; then yes, there will be a slight interuption when the partner takes over the paths.