This will discuss the share storage of iscsi in RoseHA cluster.When I configured the service of the volume settings, I add the path and mount point in the configuration tool:
![]()
As we can see above, I configured multipath service and LVM for this lun. So it came across with the multipath service and lvm service , and of course the iscsi .
When I tried to bring the database service, it can’t mount the lun and throw out many errors on I/O. that’s tricky.
![]()
![]()
Every time when I run the script it generates this error.
As we know, iscsi storage needs to first be DISCOVERED, then LOGIN the target. So if we use iscsi for shared storage, we should bear in mind of this. So I prefer to FC storage not only for performance but also for ease of deploying.
It says I can’t mount the lun:
![]()
For RoseHA cluster, if you want to switch the service to another server, then you should first mount the shared lun, then the script will start the service.
My problem on this issue is that every time when I switch the service, it didn’t first mount the lun, but run the start script, I put the DISCOVER and LOGIN command in this script, so when the cluster can’t determine the shared lun with the monitor script, of course, every time it runs the stop script, so never made it successful.
After knowing this, I know I have to make some changes on the strategy on mounting the shared lun.
Firstly, delete the “volume setting” entry, make the cluster not to check the shared lun when it check the resource when take over cluster service, but put it in the script,
Let the script itself to manage such resource as IP resource, shared lun resource, script resource etc.
![]()
In the start script, I first discover and login the target, then to restart the multipath service, and followed with scan the lvms, and active the lun, finally , I mount it here but not in the following script. This works out!
With FC storage, I could get rid of so many troubles!
