Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
Hi,
I have Informix as a source and I want to execute a Full Load, but I notice that there is a restriction: 'All transactions should be manually stopped before starting a Full Load task. Transactions started during the Full Load operation will be cached.'
Since I cannot stop transactions on the source database, what are the recommendations?
thanks.
Hello @lguevara ,
Feel free to let us know whether the above workaround meets your needs.
If not, here are a couple of alternative options:
Option 1 – Have Qlik Replicate wait when locks are encountered
In Informix source endpoint adding an internal parameter lockWait (default is 0), setting it to a reasonable value (for example, 120 seconds). This allows Replicate to wait 2 minutes for open transactions to close before proceeding, if hits locks.
Option 2 – Allow dirty reads
Enable dirty reads on the source (IBM Informix). This may result in reading uncommitted or inconsistent data so it is generally not recommended.
Regards,
John.
Hello @lguevara ,
This appears to be a timing issue. For example, when you start a task, it may take around 5 seconds to reach the RUNNING state. If there are no open transactions during that window, the startup should complete without issues.
If it’s difficult to find a quiet window during business hours, consider starting the task during off-peak periods instead.
Hope this helps.
John.
Hello @lguevara ,
Feel free to let us know whether the above workaround meets your needs.
If not, here are a couple of alternative options:
Option 1 – Have Qlik Replicate wait when locks are encountered
In Informix source endpoint adding an internal parameter lockWait (default is 0), setting it to a reasonable value (for example, 120 seconds). This allows Replicate to wait 2 minutes for open transactions to close before proceeding, if hits locks.
Option 2 – Allow dirty reads
Enable dirty reads on the source (IBM Informix). This may result in reading uncommitted or inconsistent data so it is generally not recommended.
Regards,
John.