System information
Troubleshooting
6-8
3. Host computer reports I/O errors during path/controller failover
If you use MPIO against path or controller failure, it is essential to check if your MPIO driver is installed properly. Below
are the checking items you need to do:
• A virtual disk has been mapped to LUNs of all host-interface ports in simple storage presentation configuration. Or,
a virtual disk has been mapped to a host or host group in symmetric storage presentation configuration.
• Because a LUN is not accessible during regular initialization. Install the MPIO driver after the regular initialization of
a virtual disk is done, or use background initialization.
• All cables are connected and the corresponding paths are displayed by the MPIO software utility.
• Check the device nodes from the operating system disk management utility to make sure the MPIO devices have
been created. Make sure the number of MPIO devices matches your cabling topology (see 5.1 Multi-Path IO
Solutions on page 5-1 for how to calculate the number).
4. The replacement controller cannot work for controller fail back
The replacement controller stops boot-up if it fails to discover the same set of expansion chassis of the surviving
controller. You need to make sure the expansion chassis are properly attached to the expansion port of the replacement
controller and there is no broken connection between the expansion chassis. Please also note that the I/O access and
background tasks can be migrated to the replacement controller only after the host-interfaces of the replacement
controller are properly connected to the host computers.
5. It takes too much time for path or controller failover with MPIO driver
The MPIO driver detects path failure by checking the HBA driver to get the link status. The setting of HBA determines
how much time it takes for the MPIO driver to detect path failure and to do path failover. If your controller failover/failback
is supported by MPIO drivers, it also determines how much time it takes for controller failover.
• Qlogic FC HBA BIOS Utility:
Fast!UTIL > Advanced Adapter Settings > Link Down Timeout
Fast!UTIL > Advanced Adapter Settings > Port Down Retry Count
• LSI FC HBA LSI command-line Utility: LSIUtil > Change FC Port settings > Initiator Device Timeout
Contact your HBA vendor for more information about these settings.
6. Linux pauses for a while during boot-up with dual-controller system
Because it takes time for Linux to test the standby LUNs of a controller, you might experience long boot-up time of Linux,
but it hurts nothing after the Linux is running. If you want to reduce the boot-up time, please follow the procedures below
after completing the installation of multi-path driver:
• Edit /etc/modprobe.conf (RedHat) or /etc/modprobe.conf.local (SUSE 10) to add this line:
options scsi_mod dev_flags=Accusys:ACS92102:0x1000.
• Change to the /boot directory, and build the ramdisk image with the following command: mkinitrd -f initrd-
2.6.[kernel_version].img [kernel_version] (RedHat), or mkinitrd (SUSE 10)
• Reboot the Linux system
7. Constant path failover or many disk timeout errors reported at host
Under heavy I/O loading, the operating systems might experience long I/O response time and falsely report I/O error
when the response time is over its I/O timeout value. The MPIO driver might also be informed of I/O or path failure and
the preferred I/O path would be changed constantly. This would severely hurt the performance. Extending the I/O timeout
value of your OS or applications can mitigate this problem. For example, on Windows system, it is advised to set the
registry key below to 60 or 90 seconds:
HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Disk\TimeOutValue.