Problem
If an outage effects all plexes of a mirror, it may be necessary to manually review each plex to determine which contains the most recent updates before resynchronizing. This ensures that the most up-to-date data is preserved.
Solution
This article is a part of a set on troubleshooting failed, or failing, disks. Click here to start at the beginning: https://www.veritas.com/docs/000080622 |
Table of Contents
1. Dismount and stop the volume
2. Use vxplex to dissociate all but one plex
3. Restart and mount the volume
4. Review the contents of the volume
5. Resynchronize the plexes, using the preferred copy of the data
1. Dismount and stop the volume
(Back to top)
Gracefully offline any processes that are using the volume and dismount it. After the volume is dismounted, use vxvol to stop it.
Note: If this is a Veritas Cluster (VCS) environment, offline the cluster resources down to, and including, the Mount resource.
Figure 1 - Using vxvol to stop the volume
Syntax: vxvol -g <disk_group> stop <volume> Example, with typical output: # vxvol -g proddg stop prodvol Vxprint now shows that the volume is disabled. # vxprint -g proddg -vp |
2. Use vxplex to dissociate all but one plex
(Back to top)
Vxplex can be used to either attach or dissociate ("break off") a mirror plex.
- If multiple plexes are already attached, use vxplex to dissociate all but one of them (Figure 2).
- If the desired plex is not currently attached to the volume, use vxplex to attach it (Figure 3).
Figure 2 - Using vxplex to dissociate a mirror plex (break the mirror)
Syntax: vxplex -g <disk_group> dis <plex> Examples, with typical output: In this example, vxprint initially shows that a volume named "prodvol" has three plexes associated with it. After using vxplex to dissociate two plexes, vxprint shows only one plex associated with the volume. # vxprint -g proddg -vp |
Figure 3 - Using vxplex to attach a mirror plex
Syntax: vxplex -g <disk_group> att <vol> <plex>
Note: Do not attach more than one plex to the volume while the volume is enabled. Doing so will trigger a resynchronization.
Examples, with typical output: In this example, vxprint initially shows that no plexes are associated with the volume named "prodvol." Vxplex is then used to attach a single plex with the volume. # vxprint -g proddg -vp |
3. Restart and mount the volume
(Back to top)
Use vxvol to start the volume and then mount it. By default, the device path to the block device for the volume is /dev/vx/dsk/<disk_group>/<volume>.
Note: If this is a Veritas Cluster (VCS) environment, bring the Mount resource online to mount the volume.
Figure 4 - Using vxvol to start the volume
Syntax: vxvol -g <disk_group> -fa start <volume> Example, with typical output: # vxvol -g proddg -fa start prodvol Vxprint now shows that the volume is enabled. # vxprint -g proddg -vp |
4. Review the contents of the volume
(Back to top)
Review the contents of the volume. In particular, note the most recent updates. This information can be compared with other plexes to determine which one contains the most recent data.
Once you have identified the most up-to-date plex, use vxvol and vxplex to associate only this plex to the volume (if it is not already). This can be done by using the same syntax that is described in Steps 1 through 3.
5. Resynchronize the plexes, using the preferred copy of the data
(Back to top)
After the desired plex is reattached, and the volume is online and mounted, use vxplex to attach the other plexes (Figure 3). This will trigger a resynchronization. The data that is contained within the plex that was attached first will be used to overwrite the other plexes that are attached subsequently.