pnp4nagios graphs are slow

[timchin]

After upgrading to 2012R2.4, my pnp4nagios graphs load really slowly. To the point of where I wish I hadn't committed my VM snapshot and could rollback. I restarted the server, but it had no effect.

Has anyone else ran into this issue and/or know how to resolve it? Thank you!
Tim Chin

[slansing]

Have you noticed any other strange issues? When you open a page showing your pnp graphs can you tail your apache log and share with us?

Code: Select all

tail -f /var/log/httpd/error_log

Stop the tail after they load and send it to us, thanks!

[timchin]

This is all that gets logged when bringing up a performance graph

[Fri Oct 04 14:46:22 2013] [error] [client 127.0.0.1] access to /usr/local/nagiosxi/html/backend/ failed, reason: SSL connection required

Tim Chin

[timchin]

No other strange issues. That error I posted goes back to before the update, so it's probably unrelated

Tim Chin

[abrist]

1. Have you created a netmount for the perfdata directories?
2. If you run "top" and then try to browse a graph on the frontend, do you notice high io wait, ram usage, or load on the server while it is trying to draw the graphs?

[Fri Oct 04 14:46:22 2013] [error] [client 127.0.0.1] access to /usr/local/nagiosxi/html/backend/ failed, reason: SSL connection required

You may have a half-baked ssl implementation. Can you check the steps in the following document:
http://assets.nagios.com/downloads/nagi ... ios_XI.pdf
3. What is the output of:

Code: Select all

grep use_https /usr/local/nagiosxi/html/config.inc.php

[timchin]

Thanks for the reply!

1). Nope. They are local. I do have a symbolic link to access them from https://server.domain.com/nagios/perfgraphs in addition to https://server.domain.com/nagiosxi/perfgraphs. It's slow from both URLs, though, and never used to be for either until the last upgrade of NagiosXI.

2). During the load of a view with 8 graphs, CPU load is about 50%, io wait is normally 0 (saw it go to 3%), and memory usage stays right around 70% used. I have a VM with 2 CPU, 768MB RAM, and 8GB disk. It was the pre-built VM template from when I purchased NagiosXI. 72 hosts / 650 checks (mainly NRPE).

3). It was false, so I changed it to true and restarted httpd & nagios services. No change. All other portions seem to be correct and they were working find before the last upgrade of NagiosXI.

Thanks!
Tim

[slansing]

Do you have the resources to double the core and memory count? And expand the disk for future proofing? This is what we recommend for your system:

http://assets.nagios.com/downloads/nagi ... ements.pdf

Are you getting any errors or hitting any limits when you tail your system log and open a slow page?

Code: Select all

tail -f /var/log/messages

[SDK]

We have the same problem in our environment since the update to 2.5

The pnp4nagios graphs take extremly long to load.

I noticed that when the graphs are drawn, the rrdtool process spikes on the server (CPU).
I don't think it did before the update.

Kind regards

SDK

[slansing]

Please answer the questions above. I have not ran into any of this on my local XI systems here which are running 2012 2.5. How are you reproducing this? Which specific pages are you visiting which load the graphs slow?

[timchin]

I went to 4 CPUs and 2GB of RAM. No difference at all.

Nothing shows up in /var/log/messages when loading an 8 graph view