Stopping and Starting iSCSI Services
Should it become necessary to manually stop the iSCSI services in Linux,
certain steps must be followed to maintain parallel processing between the
RAID storage array and the host server. MPP (multi-path driver) is partially
unloaded when the iSCSI services are stopped.
To ensure orderly shutdown of the iSCSI services, take the following actions:
•
Stop all I/O.
•
Unmount all correlated filesystems.
•
Unload all modules in correct order.
Internally, these tasks can be handled by the mppiscsi_umountall script, which:
Kills all pending I/O on each iSCSI-mounted devices via the fuser
•
command
•
Unmounts all iSCSI-mounted devices
•
Unloads the mpp_Vhba module
•
Unloads the iSCSI module
Likewise, to ensure orderly startup of the iSCSI services, these actions must
occur in the following order:
1 Load the mpp_Vhba module (if not already).
2 Load the iSCSI modules.
NOTE:
Unloading the iSCSI modules will kill all I/O on the iSCSI devices.
RAID Level Migration Requirements
Successful RAID Level Migration (RLM) requires that the appropriate
number of physical disks are present in your disk group. To perform RLM
from a RAID 1 or RAID 10 to an odd-number of physical disks (five or more)
in RAID 5, first add an even number of disks to the RAID 1 or RAID 10,
convert to RAID 5, then add your odd physical disk.
6
Information Update
All manuals and user guides at all-guides.com