Broadcast screen slow to open

Bogdan Kiselitsa shared this problem 11 months ago
Resolved

Hi,


We are experiencing an issue with there Broadcast screen for a report can take a very long time to open. It seems to be proportional to the number of broadcast schedules configured and the number of historical events.


Looking at the server logs, I've traced it back to a query that seems to run for each Broadcast schedule:

SELECT EventId, EventTypeCode, GMTDateTime, IpSource,
   SessionId, FinancialTransactionValue, FinancialCommission,
   FinancialDestinationValue, FinancialTaxValue,
   EventTime, EventDate, EventCode, EventData, ReferenceId,
   UnitId, TomcatSessionId, ReasonCode, ReasonDescription, ContentId
FROM EventArchive
WHERE GMTDateTime >= 20171209013729
AND GMTDateTime <= 20180123022156
AND EventTypeCode != 'SYSTEMTASK'
ORDER BY EventId;

In our case this can take a while to run, and returns 184,000 rows!


We are seeing this on 7.4 build 20180208


Is there any fix for this issue forthcoming?


Regards,

Bogdan.

Comments (12)

photo
1

Hi Bogdan,

Thank you for reaching out. Returning 184,000 rows actually shouldn't be causing a noticeable slowdown. I think a good place to start would be if you could append /info_cache.jsp to your Yellowfin server name in your browser, scroll down to the bottom and take a screenshot of that page:

b43c6c12db2188e1bd68a3d1cbf8d4f5

Thanks,

Mike

photo
1

Hi Mike,


Thanks for that.


I've had a look at ours (below), and it seems like we get a lot of misses, and the period cached is too small.


I also came across your KB article here: https://community.yellowfinbi.com/knowledge-base/article/what-is-event-caching-and-how-can-i-adjust-it


So we will implement some of these changes and see if it resolves the problem.


Thanks.


e9642c683a02797aa799d95c62cdb503

photo
1

Hi Bogdan,

Way to proactive! Yes, please let me know how that works out. I look forward to your response.

Regards,

Mike

photo
1

Hi Mike,


The changes to the event cache have definitely improved the run time, and now we are getting really high hit rates.


However, since we made the change, it looks like the run visualiser on each broadcast is not updating correctly?


If you look at the below image, we are not seeing any runs since Wed the 14ths, when we made this change, and I've verified that broadcasts are in face being sent out.

Could it be that now we are seeing outdated (cached) data here?

c2b462ce73ab8ade5fcc8a5052e8676b

photo
1

Hi Bogdan,

Thank you for your response. I think what you may be looking for can be found on our Wiki page here. You can set an automatic data cache refresh on your data store in your view to run every x minutes, hours, days, weeks, etc. Why don't you give this a try and let me know what you come up with here?

Thanks,

Mike

photo
1

Hi Mike,


I think view caching (which I wasn't aware of, thanks!) is not applicable here, as this is a system feature. It's not a view we created, but rather the out-of-the-box display of schedule broadcasts, so I don't have control over it.


Regards,

Bogdan.

photo
1

Hi Bogdan,

Thanks for your response. I'm wondering if, even though you bumped it, whether your Events Cache is at 100% again. The run visualizer pulls exclusively from cached data, so it seems there's a good possibility that on the 14th you hit 100% cached and so it hasn't since updated on the visualizer since it may not have cached any additional data since then. Can you send your updated info_cache.jsp information?

Regards,

Mike

photo
1

Hi Mike,


Our current stats look like the below.

But even if the cache is full, it shouldn't display old data. Do we need to raise this as a bug?


6390824360a5b40a0dbb523c3d2f67ed

photo
1

Hi Bogdan,

I think you are right. This seems like some unusual behavior. Unfortunately, I think our next steps here would be if you could send us your Config DB and a SQL query that supplies your empty tables, so we can get your data structure to investigate further and see what's going on first-hand in your instance.

Regards,

Mike

photo
1

Hi Bogdan,

I just wanted to check in and see how things are going with this. I'm also wondering if perhaps the scheduler is showing as expected now. Since the cache size was increased it may have taken some time to build that cache. Before being done building you may have still been seeing the old data there.

Regards,

Mike

photo
1

Hi Mike,

Thanks for checking in.

Yes, it seems like it's mysteriously working properly now. At least on one of our nodes the Events Cached is at 100%, if that's what made the difference.

Still seems like odd behaviour for a cache to me, but it's no longer a problem it seems.

Thanks for your help.

Regards,

Bogdan.

photo
1

Hi Bogdan,

That's good to hear. I'll go ahead and mark this as Resolved then. Please don't hesitate to reach out with further questions or concerns.

Regards,

Mike