Advice pls OES Linux cluster iSCSI drives separate NIC

Question

In order to improve backup performance I am thinking of connecting our iSCSI drives (on Synology NAS box) to the 2 OES 11 Cluster servers via their second network cards. I.e. the iSCSI drives will be on a separate network so the data is not travelling over the same network twice in order to get to the backup server (Backup Exec on Windows).
I would welcome general comments on any pros and cons of this idea
Not being a network expert I am not sure how I can do this. As I understand it the second network cannot be in the same subnet as the main network or confusion reigns. However my cluster resources will have to be in the main subnet range. Is it that the iSCSI portal addresses can be on a different subnet but everything else is on the main subnet? Will I have any issues if I change the IP addresses of the iSCSI Initiators?
Thanks in advance

Answer 1

Originally Posted by NickHoare
In order to improve backup performance I am thinking of connecting our iSCSI drives (on Synology NAS box) to the 2 OES 11 Cluster servers via their second network cards. I.e. the iSCSI drives will be on a separate network so the data is not travelling over the same network twice in order to get to the backup server (Backup Exec on Windows).
I would welcome general comments on any pros and cons of this idea
Not being a network expert I am not sure how I can do this. As I understand it the second network cannot be in the same subnet as the main network or confusion reigns. However my cluster resources will have to be in the main subnet range. Is it that the iSCSI portal addresses can be on a different subnet but everything else is on the main subnet? Will I have any issues if I change the IP addresses of the iSCSI Initiators?
Thanks in advance
Since searching for cluster advice regarding iSCSI brings up this post near the top (!) I thought it worthwhile asking this secondary question to make this thread more of a reference.
Looking at previous OES Netware guide to using iSCSI in a network (8 years old but the closest thing to what I was looking for) it suggests that the iSCSI target (I think I mean initiator here - volumes are on a NAS box) is set up as a cluster resource with associated load and unload scripts. The documentation for iSCSI implies, but does not state clearly, that the iSCSI initiators should all be set up and connected beforehand. Presumably the assumption is that automatically or onboot connected initiators would already be connected.
Seeing as I hope not to have to do much with the system for weeks on end I am concerned that something could happen to the Nas box/network etc whilst the servers remained up so that connections would be lost at some point. Come the failover the iSCSI targets would not be there so all pool resources would fail to connect.
To that end I want to add, and I think as separate resource, a script to attempt to connect the iSCSI initiators in case they are not connected but not to fail if they already are. My proposal is this
#!/bin/bash
. /opt/novell/ncs/lib/ncsfuncs
ignore_error iscsiadm -m node --loginall=automatic
exit 0
which seems to work. In our simple 2 node network I may as well connect all with the advantage that the script does not need to be changed if we add further pools. (we have 2 pools each with 2 volumes which goes against some unexplained 'best practice' notes in the documentation). No point in testing for failures which might be hard anyway as there is presumably an error if they are already connected.
The unload script just has exit 0 .
Any comments or issues I should be aware of?