Capacity Planning Report Performance

[andyb4u]

Hi,

I've been tasked to see if there is any way to improve the performance of a capacity report we are running.

Does the capacity planning report gets it's data straight from the database or does it need to look through the /usr/local/nagios/var/archives/ folder at all?

My colleague ran a capacity planning report against a host group, which has 31 servers in it for the next month. From start to finish it took around 30 minutes and resulted in the actual NagiosXI server going into a critical state as all RAM was consumed with only 337 MB left. The generated PDF was 368 pages long.

Is there anything we can do to speed the capacity planning report up?

I have tried testing a smaller hostgroup and can see that the wkhtmltopdf process consumes a large amount of memory (GBs) to produce the PDF.

PHP settings are currently:

max_input_vars = 10000
memory_limit = 512M
max_execution_time = 180
max_input_time = 180

Our DB is offloaded.

Regards,
Andy

[lmiltchev]

Does the capacity planning report gets it's data straight from the database or does it need to look through the /usr/local/nagios/var/archives/ folder at all?

The data is obtained from the DB via a python script.

My colleague ran a capacity planning report against a host group, which has 31 servers in it for the next month. From start to finish it took around 30 minutes and resulted in the actual NagiosXI server going into a critical state as all RAM was consumed with only 337 MB left. The generated PDF was 368 pages long.

Most likely the issue is that wkhtmltopdf has to hold onto the RAM memory while it runs each individual prediction. It's surprising that the report even finished running... We don't recommend running capacity planning on that many objects.

Is there anything we can do to speed the capacity planning report up?

Not really. This is a common issue and we've thought about making a limit to how many objects we should be showing on the capacity planning page/pdf. Break up your hostgroup into smaller "sub-groups", and run the report against smaller number of objects. Instead of one very large report, you could have a few smaller reports.

[andyb4u]

Thank you for answering my questions so promptly. It's much appreciated.

Just one more question that I'm sure I'll get asked - in the roadmap for XI 6 it says: New report management system and backend report generation (running reports in the background) including saving reports.

Is it likely that the capacity planning report will run quicker against larger number of objects or will it still run into the same issue when producing the pdf (using wkhtmltopdf)?

[lmiltchev]

The Capacity Planning report won't be much faster but it'd be able to run in the background, meaning you could still use the system fine while it's running. What I learned from our developers is that Nagios XI would be using something like nice to keep it from pegging the CPU. I don't have any other technical details at the moment.

Let us know if you have any more questions or it is OK to close this topic. Thank you!

[andyb4u]

Thanks again for your replies. You can close the topic.