| Sun StorageTek Availability Suite Troubleshooting Guide |    
|
| Sun StorageTek Availability Suite Troubleshooting Guide |
|
| C H A P T E R 5 |
|
Point-in-Time Copy and Remote Mirror Interoperability |
This chapter describes Point-in-Time Copy and Remote Mirror software interoperability issues.
The following topics are included:
The term ndr_ii is used to define a mechanism within the Remote Mirror product set to use a Point-in-Time Copy volume to assure that there is always a write-order consistent copy of a replicated volume on the secondary node.
During normal Remote Mirror replication, the secondary volume is kept write-order consistent with the primary volume, even under conditions of brief link failures or data link latency. As long as there is ample room to store the ordered write I/Os into either the memory-based or disk-based queues, write-order consistency is maintained. When there is no longer room to queue write I/Os (or through operator intervention), a Remote Mirror set will be placed into logging mode and the ability to maintain write-order consistency is lost. Note that even at this point, the secondary volume is still write-order consistent, although no new write I/Os are being replicated.
To resume write-order replication, either through operator intervention or by enabling auto-sync capability, an update synchronization operation must first be performed. It is during this period of time, update synchronization, that write-order would be lost if an ndr_ii pair is not enabled and used. Prior to the start of the update synchronization operation, an automatic Point-in-Time Copy of the write-order consistent secondary volume is taken. Then the sequentially-ordered update synchronization is performed and, if successful, the Point-in-Time Copy is deleted, since the secondary volume is now write-order consistent. In the unlikely, though possible, event of a failure during the update synchronization operation, the Point-in-Time Copy shadow volume can be used to restore a write-order consistent volume back on the secondary node.
Issues with ndr_ii pairs are most commonly seen when a system administrator has configured an ndr_ii pair, yet when it comes time to use the ndr_ii pair, the Point-in-Time Copy set fails to configure. To assure correct behavior:
1. Prior to configuring an ndr_ii pair, the system administrator should configure the Point-in-Time Copy set.
Use the Remote Mirror secondary volume as the Point-in-Time Copy Master volume, along with a (compact) dependent shadow volume and correctly sized bitmap volume.
3. Manually force the Remote Mirror set into logging mode and then back into replicating mode.
The ndr_ii mechanism should recognize and use the already configured Point-in-Time Copy set.
4. After the Remote Mirror set has completed synchronization, check for errors on both the Remote Mirror and Point-in-Copy volumes.
The command to complete synchronization is sndradm -w [set].
5. If this test completes correctly without error, the Point-in-Time Copy set can be disabled, then the test where Remote Mirror is set into logging mode and then back into replicating mode can be repeated.
During this test the Point-in-Time Copy set will be temporarily enabled and disabled automatically.
| Sun StorageTek Availability Suite Troubleshooting Guide | 819-6151-10 |
|
Copyright © 2006, Sun Microsystems, Inc. All Rights Reserved.