centreon / centreon-collect

Centreon collect software collection
10 stars 5 forks source link

Step in RRDs does not match check interval #797

Open cgagnaire opened 1 year ago

cgagnaire commented 1 year ago

BUG REPORT INFORMATION

Prerequisites

Versions

For the RPM based systems

centreon-broker-storage-22.10.0-16.el8.x86_64 centreon-broker-cbd-22.10.0-16.el8.x86_64 centreon-broker-22.10.0-16.el8.x86_64 centreon-broker-cbmod-22.10.0-16.el8.x86_64 centreon-broker-core-22.10.0-16.el8.x86_64 centreon-engine-22.10.0-16.el8.x86_64 centreon-engine-daemon-22.10.0-16.el8.x86_64 centreon-engine-extcommands-22.10.0-16.el8.x86_64

Operating System

Oracle Linux 8

How the the component has been installed and versions

Description

It looks like RRDs are not created with the right step. When I push new conf, after the first check that retrieve metrics, the RRDs are created with a 5min step eventhough the check is not configured to be executed every 5min. On the other hand, when I rebuild the graph, the step correspond with the check interval.

Steps to Reproduce

  1. Create a check with a 1min interval,
  2. Dump the RRD file (rrdtool dump /var/lib/centreon/metrics/.rrd .xml),
  3. Rebuild the RRD (through UI),
  4. Dump the RRD file.

Describe the received result

First case : the pdp_per_row of the first rra header will be 300 (5min) <pdp_per_row>300</pdp_per_row> <!-- 300 seconds -->

After rebuild : pdp_per_row is 60 (1min) <pdp_per_row>60</pdp_per_row> <!-- 60 seconds -->

Describe the expected result

A pdp_per_row corresponding to the check interval at RRD creation.

Additional relevant information (e.g. frequency, ...)

After saying all this, it looks like it is working correctly sometimes :) But I can't say why... after a reload or a restart, I don't know...

cgagnaire commented 1 year ago

I had a case where I push several services with different check interval. 3 with a 1min interval, 4 with a 5min interval. One graph respect the 1min step, the others are 5min.

bouda1 commented 11 months ago

Hi @cgagnaire

We have worked on this issue recently. I hope the fix will be available very soon. Cheers.