Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
Environment:
QV11 SR2
Question:
In QV11 the Publisher FAQ states that task reload and distribution are handled in two different steps, each requiring its own QV engine. I have many (hundreds) of QV tasks building QVDs, and I am noticing that for a given task the reload will complete, release its QV engine, and then it gets stuck for 20+ minutes waiting for a QV engine to become available to start the distribution task (see attached TaskLog). The catch is that once an engine is available the distribution portion takes mere seconds to complete.
The task is building a QVD, not a user facing QVW app – so why do I need to wait for ‘distribution’? Is there a way for me to configure the task to tell QVDS to only run the ‘reload’ portion and not the ‘distribution’ portion? For the task in question in the QMC under the distribute tab all settings are blank (see attached picture). Any ideas?
Links:
QV 11 DS FAQ:
Thanks in advance!
Trevor
For future readers: I worked with Qlik support to switch to "Distribute to Folder". I found that when you configure a task to "Distribute to QlikView Server" it does occupy two engines - one to refresh and a second one to distribute. If you utilize "Distribute to Folder" you only use a single engine to refresh. This solved my original issue and did seem to work well for us.
One down side, however, is impact on future system upgrades and migrations. There are situations (upgrades/migrations) that would require you to manually adjust each "Distribute to Folder" Path property. We used this method for a few years and this problem of hard-coding the distribution to folder path came up twice in the context of upgrades/migrations (we had several hundred tasks and would have needed to modify about half of them manually). We have since switched back to "Distribute to QlikView Server" so we can be in a better position for future upgrades (our original challenge with engines has been resolved with environment upgrades).
I realize QV11 included a new ‘Build QVD feature’ where you add a supporting task to build QVDs only via script. The problem is that I am building 300-400 QVDs and they are all built using QVWs now, so this would be a very inconvenient fix as it would be a ton of administrative work.
QT Support stated today that QV Engines are assigned based on the start time of the task. So, if Task A has just finished the reload phase, releases its engine, enters into the distribution phase and is seeking a QlikView engine – one would be assigned based on the start time of the task. This should mean that this original task should get priority over any task that is chronologically ‘later’ (which is a good thing).
Hi Roth,
As you don't need to distribute the report and want to do the reload only , than in this case you can create a External task , which pass only reload instruction to the QMC.
"qv.exe file path\qv.exe" \r "application path\application_name.qvw"
Pass the above command while creating the external task.
NOTE:- In this you won't get the task log in QMC.
Thanks,
D J
Thanks for your input Dhiraj Kumar. I would like to avoid adjusting any tasks because I have such a large quantity of them in existence. You have a great point about an external task not using distribution, so thanks for your input.
- Trevor
For future readers: I worked with Qlik support to switch to "Distribute to Folder". I found that when you configure a task to "Distribute to QlikView Server" it does occupy two engines - one to refresh and a second one to distribute. If you utilize "Distribute to Folder" you only use a single engine to refresh. This solved my original issue and did seem to work well for us.
One down side, however, is impact on future system upgrades and migrations. There are situations (upgrades/migrations) that would require you to manually adjust each "Distribute to Folder" Path property. We used this method for a few years and this problem of hard-coding the distribution to folder path came up twice in the context of upgrades/migrations (we had several hundred tasks and would have needed to modify about half of them manually). We have since switched back to "Distribute to QlikView Server" so we can be in a better position for future upgrades (our original challenge with engines has been resolved with environment upgrades).