PRIMECLUSTER Global File Services Configuration and Administration Guide 4.1 (Linux)

C.5 Action to be taken when the file system configuration information is inconsistent

This section explains procedure when it becomes impossible to mount the GFS Shared File System because super block or partition configuration data differ from each copy data.

Perform the recovery in accordance with the following procedures.

1. Recover the file system configuration information.

# sfcadm -r /dev/sfdsk/gfs/dsk/volume1

2. Recover the consistency of the file system.

# sfcfsck /dev/sfdsk/gfs/dsk/volume1
  :
# sfcfsck -o f,nolog /dev/sfdsk/gfs/dsk/volume1
  :
#

3. Mount the recovered file system.

# sfcmntgl /dev/sfdsk/gfs/dsk/volume1

4. Confirm that the recovered file system can be mounted on all nodes sharing GFS Shared File System.

# mount

5. Confirm on all nodes that the following messages are not output in /var/log/messages when sfcmntgl was performed in step 3. No need the following operations if the message is not output.

WARNING: sfcfs_mds:2042: fsid: The size of the partition might have been changed after the file system was made.

If the message is output, there is a possibility that the GDS volume used for the GFS Shared File system was expanded. The file system can keep on using, but please perform the following operations promptly because GFS does not deal with expanded volume of GDS.

6. Back up the files in the recovered file system. For the way to back up, refer to "Backing up file by file".

Please be sure to perform the backup with each file.

7. Unmount the recovered file system.

# sfcumntgl /dev/sfdsk/gfs/dsk/volume1

8. Confirm that the recovered file system can be unmounted on all nodes sharing GFS Shared File System.

# mount

9. The parameter information on a GFS Shared File System is evacuated.

# sfcmks -m /dev/sfdsk/gfs/dsk/volume1 > /tmp/mkfs_cmd

The command line used when the file system was created is stored in /tmp/mkfs_cmd by the above procedure.

10. Delete the recovered file system.

# sfcadm -D /dev/sfdsk/gfs/dsk/volume1

11. Confirm the file system is deleted.

# sfcinfo -a

12. Depending on the situation, re-create the partition.

Please re-create GDS volume only when you will return to the volume size before expanding GDS volume and resume manipulation.

It is not necessary to re-create if resume manipulation in the volume size after expanding GDS volume.

Please refer to "operation" in the "PRIMECLUSTER Global Disk Services Configuration and Administration Guide" about the operation method of GDS volume.

13. Create the GFS Shared File System using the same partition or the partition created in step 12.

# sfcmkfs -o node=sunny,moony /dev/sfdsk/gfs/dsk/volume1

When creating the GFS Shared File System with the same parameters that used before volume expansion of GDS is performed, it is created using the information which evacuated in step 9.

# sh /tmp/mkfs_cmd

14. Confirm that the GFS Shared File System is created.

# sfcinfo -a

15. Mount the created file system.

# sfcmntgl /dev/sfdsk/gfs/dsk/volume1

16. Restore the file backed up in step 6. For the way to restore, refer to "File-by-file restoring".