Forum

This content is now out of date.

Visit Our Community

Yellofin 7.1 repository performance with MySQL

Hello,

We are currently running a 6.3 and 7.1 version of Yellowfin in parallel and notice quite a difference in the performance of the repositories, which are running on the same MySQL server. E.g. opening a report in edit mode takes seconds in 6.3 and minutes in 7.1.

I just want to check, if there are any known issues that might explain this behaviour or if you have any idea what the cause could be or even better a solution or way to pinpoint the problem?

Many thanks in advance and best regards,
Ingo
Hello Ingo,

Thank you for posting your question / issue on the forum.

As you are likely aware, there are many, many factors that come into
play when opening or running a report. Can you confirm that the
report you are testing with is the same report in both cases (6.3 & 7.1)?

I think the real issue to focus on is why this report is taking minutes to load
in your 7.1 environment?

- Which Browser are you using? Have you tested using different ones for comparison?
- What is the makeup of the report? Crosstab? Cached Filters? Calculated Fields?
- Is the poor performance happening with all reports?
- Is the view that these reports are based on fairly complex? If so, do you have Optimize View enabled?
- What are some of the specifics of your environment? What is the current YF DB size?

-Can you send across the Info_threads results when experiencing the slow performance, it will show exactly what YF is doing at that very point in time.
This can be accessed by opening a new browser tab and adding it to YF URL. E.g :YellowfinServer:8080/info_threads.jsp .
The more results, the better, but it does depend on the area.

I've also included an attachment to our latest server spec guide.

Please get back to us on the above and we will continue to look at this for you.

Regards,

Kyle



Hi Kyle,

Thanks for the input. I send some files by email to the support address.

Best regards,
Ingo
Hello Ingo,

I'm just letting you know that we received the logs in
support. Thank you for sending those along.

Once we've gone through them we will respond to this
forum thread.

Thank you for your continued patience.

Regards,

Kyle
Hello Ingo,

Just circling back with you on your query.

In Yellowfin 7.0 + , when a report is edited, it�s actually �cloned� so that if you make changes to the report, you can roll it back.
This means every report edit, is also copying the report, so it can take longer.
Part of this cloning process is re-caching the filters, which can really slow things down.

However, we recently made a code change that stopped the filters from re-caching during this clone process.

Can you confirm which build of 7.1 you are currently testing with?

Unfortunately, the info threads didn't shed too much light on why the report is loading so slowly.

Is it possible to address some of the questions outlined in my initial email?

- make up of report?
- browser you are testing with? Have you tried different browsers?
- is this happening with all reports?

We look forward to your response.

Thank you,

Kyle
Hi Kyle,

This was a very good hint. Indeed some of the reports, and the one that is the base for this thread, have cached filters which take a long to populate. I upgraded the instance and saw some promising first results.

When we went into debug mode, we noticed in the log file that the filter caches seem to use some old schema names and data sources, which were either changed or deleted in the meantime since they were first created.

How can I see and change the data sources and statements which are used when populating/refreshing the filter caches?

Many thanks and regards,
Ingo
Hi Ingo,

Thank you for updating us on your progress.

I will look into your last question today and attempt to
provide you with some useful information.

Cheers,

Kyle