Troubleshooting SnapSync Job History
To display the job history, navigate to Delphix Management > Manage > Dashboard.
To find all of the SnapSync (a.k.a. DB_SYNC) jobs for that dSource, enter a phrase like SnapSync for database "XYZ", where XYZ is the name of the dSource in the field Filter By Keyword on the Job History.
The start and end times for the job will be shown in each entry.
- If the job history shows that the SnapSync jobs have had approximately the elapsed time duration as always, this indicates that nothing has changed recently, which helps in trying to understand the nature of the problem.
- If the SnapSync job durations have always been unsatisfactory, you need to further examine how the dSource was setup.
- If the job history shows that the SnapSync jobs are taking progressively longer amounts of elapsed time, this may indicate that the source database is simply growing progressively larger or busier.
- If the job history shows a sudden change in the duration of the SnapSync jobs, this indicates that something may have changed in either the configuration of the Delphix dSource, or that something may have changed in the configuration of the source database or the source environment (i.e. database host/server/cluster).
Measuring Daily Change Rate in an Oracle Database
If you are using RMAN, look at the size of the (daily) incremental dumps. This information is stored in the DB Control File or in the RMAN Catalog.
You may have to look at several time periods (e.g., one week, one month, one quarter) to calculate the average daily change rate.
To calculate the average Daily Change Rate see the SQL script:
select to_char(completion_time, 'DD-MON-YY') as "Day", round(sum(blocks * block_size)/1024/1024) as "Data Change (MB)", to_char(round(avg(blocks*100/datafile_blocks)), '90.99') as "Percent Data Change" from v$backup_datafile bdf where nvl(incremental_level, 0) >= 1 group by to_char(completion_time, 'DD-MON-YY') order by to_date(to_char(completion_time, 'DD-MON-YY')); Day Data Change (MB) Percen --------- --------------- ------ 13-OCT-15 39738 4.00 14-OCT-15 120601 2.00 15-OCT-15 120329 2.00 16-OCT-15 116750 2.00 17-OCT-15 71035 8.00
One possible explanation for the sudden increase in SnapSync timing might be that a recovery was performed on the source database, causing the SnapSync service to perform a "link" (a.k.a. initial snapshot) of the database again.
Another possibility, specifically for Oracle dSources, is that Block Change Tracking (BCT) may have been disabled on the source database. See documentation for information on enabling BCT.
These are not the only possibilities, but knowing when sudden changes happened in the duration of the SnapSync jobs can help you locate what changed in the source database or source environment (i.e. database host/server/cluster).