Stopping Replicate tasks while Source DB still active
Client needs to do maintenance on the target DB and has asked for the Replicate tasks to be paused. If I pause the tasks, can the Source database remain active and any activity will catch up when the tasks are restarted? If so, do I need to start the tasks from a point in time or just resume?
This is last minute request for maintenance tonight and I want to be sure I manage the tasks correctly. Any assistance is appreciated.
You didn't mention what platform. But for SQL Server, as long as the databases are in Full or Bulk-log recovery, once the tasks are RESUMED (as normal, not advanced options), it will pick up where it left off. It may take a while to catch up, depending on how long the tasks were down. Also, the Transaction Log backup files should be available, since the task will try to read them. HTH
You didn't mention what platform. But for SQL Server, as long as the databases are in Full or Bulk-log recovery, once the tasks are RESUMED (as normal, not advanced options), it will pick up where it left off. It may take a while to catch up, depending on how long the tasks were down. Also, the Transaction Log backup files should be available, since the task will try to read them. HTH
What source DB? As @PGN indicated you should be fine to stop and resume as long as the redo-log or t-logs or wal files or whatever they are called for the DB in us remain available from the stop time.
Many customer have those configured to remain available for 48 hours so you should be fine. You may want to check with the source DB DBA and if need be ask them to suspend any log purging until the target is back, the task resumed and 'caught up'.
Another option, but not for 'tonight' is to consider converting to a LOGSTREAM task where Replicate 'sucks' the changes to an internal storage format on the Replicate server and applies from there. That way you have full control over the storage retention in the Replicate environment. This is extra interesting of multiple tasks take changes from the same source, but for a basic one-source to one-task to one-target setup it is admittedly a bit over the top.
Source DB is DB2 LUW. I will inquire as to how long redo-logs are available for. I think however, for this evening, until I am comfortable in what has to be done, I will take the Source database application down. Given I do not know how long maintenance will take and therefore catchup, I think for production - better to be safe.
Thank you - will inquire about the logs - pretty sure they are setup, but for tonight I think I will have the Source Application down, until I can confirm the process.
Exactly why I am pushing back 😉 - the request to pause the replication only came up this afternoon, thus why I am planning on gathering the details for a runbook and test it in the QA environment. This is a relatively new implementation (up since August) and this is the first time this has come up. Do you know if there is a place in the community with possible suggested scenarios/runbooks for similar situations?
