DCM log fails to completely drain - Veritas Storage Foundation for Windows Volume Replicator Option

Article:TECH55550  |  Created: 2007-01-15  |  Updated: 2008-01-15  |  Article URL http://www.symantec.com/docs/TECH55550
Article Type
Technical Solution

Product(s)

Environment

Problem



DCM log fails to completely drain - Veritas Storage Foundation for Windows Volume Replicator Option

Solution




Description

In most configurations, a DCM (Data Change Map) is used when an SRL (Storage Replicator Log) overflows. Unlike the SRL, which is a transaction log, the DCM log is a simple bitmap. Each "bit" in this bitmap corresponds with a region of blocks on the replicated volume. When a write occurs within a particular region of the volume, the corresponding bit is marked as "changed" (Figure 1).

Figure 1
 


It is important to understand that even if the amount of data that is written to the volume is very small, the entire region will be marked as "changed" and will need to be completely replicated again. As a result, the DCM is usually significantly less efficient than the SRL and often results in the same data being replicated more than once.

In some cases, a process that is running on the primary may periodically write data to the same region of a volume. If the next write occurs before the DCM has finished draining, the region may be marked as "changed" again. This may result in a situation where the same region must be replicated repeatedly and the DCM is unable to completely drain.

Note: During an initial synchronization, all regions are marked as "changed." The initial resynchronization is similar to cases where an SRL overflow has occurred, except that all regions are marked as "changed."


Recommendations

In most cases, the DCM will eventually drain during a period of low disk activity on on the primary. In some cases, it may be necessary to stop the applications or processes that are running on the primary, and are writing to the replicating volume. In addition, manually disabling access to the volume will completely prevent writes from occurring on the primary. This allows the DCM to drain.

WARNING: Before disabling access to a volume, verify that no open handles exist. Forcibly ending open handles will have unpredictable results, including possible data corruption.

Disabling access to a Volume

1. From VEA (Veritas Enterprise Administrator), expand Replication Network.
2. Expand the RDS (Replicated Data Set).
3. Expand the primary RVG.
4. Right-click on the replicated volume(s). Select Disable Data Access.


keywords: stuck in DCM, DCM mode,


Legacy ID



294170


Article URL http://www.symantec.com/docs/TECH55550


Terms of use for this information are found in Legal Notices