No relevant resource is found in the selected language.

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>


To have a better experience, please upgrade your IE browser.


LV Become Inaccessible After Installed the Multipathing Software

Publication Date:  2012-07-22 Views:  2 Downloads:  0
Issue Description
The LUNs on the S2600 are mapped to an HP host (running SUSE 10 SP 1). Logical volumes (LVs) are created and accessed on the host before the multipathing software is installed. However, after the multipathing software is installed on the host, the LVs become inaccessible and services are interrupted. The fault persists after the multipathing software is uninstalled.
Alarm Information
Handling Process
Step 1  Change the name of the file .cache under the /etc/lvm/ directory to .cache~. (The operation can be undone upon failures).
Step 2  Run the pvscan command again to scan physical volumes. Step 3  Run the fdisk –l command and no Input/Output error message is displayed.
The error is rectified.  
Root Cause
Step 1  Run the fdisk –l command.
The error messages /dev/dm-0 Input/Output error and /dev/dm-1 Input/Output error are displayed, indicating that dm-0 and dm-1 are inaccessible. (dm-0 and dm-1 are device IDs that are automatically generated when the two LVs are created. Their meanings are the same as sbd and sdc.)
The other disks displayed by the fidks –l command are accessible. The number of physical links displayed by the Isscsi command doubles that by the fdisk –l command. (The Isscsi command shows both physical and logical disks, while the fdisk –l command shows only logical disks.) It indicates that physical links are normal, and I/O applications to the S2600 are available.
Step 2  Run the pvdisplay, vgdisplay, and lvdisplay commands to view the volume information.
The information is displayed as normal.
Step 3  Run the pvscan, vgscan, and lvscan commands to scan physical volumes, and the operations succeed.
Therefore, the reason for the failure must not be that the volume information is lost or the volume is deactivated. However, dm-0 and dm-1 are still inaccessible.
Step 4  Run the vgchange –ay and lvchange –ay commands to reactivate the VGs and LVs, but dm-0 and dm-1 remains inaccessible.
Then the question comes forth: Why the LVs fail to be accessed when the physical links and volume information are normal?
The answer is that the LVs obtain the LVM access path from the LVM cache; however, after the multipathing software is installed, the path becomes a virtual path that is invisible to file systems, causing I/O applications fail to be processed.
There are two solutions:
1. Restart the host and then run the pvscan command. The access path is retrieved.
2. Delete the /etc/lvm/.cache file and then run the pvscan command. The access path can be retrieved based on the UUID of the PV. (After the pvscan command is executed, the LVM access path is first queried in the .cache file and then by the file system if the .cache file does not exist.)
Create PVs, VGs, and LVs after the multipathing software is installed to eliminate this error.