Overview

One major upgrade to the Plesk Obsidian release (2019) was the Advanced Monitoring extension. This added easier to follow graphs and will provide more advanced features in the future as well.

Interpreting the Graphs

CPU

While having access to the metrics is a great start, knowing how to interpret them correctly is key. Higher resource utilisation when your sites are busy is expected behaviour, which is why it’s also important to measure your website metrics separately as well (we recommend services such as Panopta). Here’s an example of a busy site:

Similarly, here’s a similar server when the sites aren’t busy:

Neither of these indicate an issue, which is why it’s important to interpret the data. The first graph with higher usage still has plenty of CPU headroom. If you hover over the graphs, you’ll see detailed metrics. For example, from the busy server:

We can see that the wait times (where the system is waiting on disk access) is at 0% so no delay and we can see that the CPU was idle for 50% which indicates it has sufficient headroom to process more information.

If you have idle times constantly lower than 50%, we recommend more CPU resources are added to your server so that you can ensure there are no delays for future traffic.

Memory

Memory usage can be a tricky one to diagnose, as servers are designed to use all available memory. This means that we need to fully consider where and how this memory is allocated. The key component we’re looking for us the used memory figure (in green). For a healthy server, this should remain below 50% of the server’s total memory, as indicated by the following graph:

We see memory also being used for file caching (yellow), which is critical to ensuring a fast website.

When there’s an issue with the amount of memory, it will generally be one of two different patterns.

Example 1: heavy usage

In example 1, we can see high amounts of memory used with very little free memory. This is an indication that the server needs more memory added (ie, purchased from Conetix.)

Example 2: Abnormal spike

In example 2, we can see that there’s an abnormal spike which indicates a potential code issue and/or recursive loop using everything available.

For both instances, this lack of memory was leading to the websites under-performing as well as displaying error messages to the end user.

We’re here to help

If you’re at all concerned by the memory usage, you can simply contact Conetix and our expert team will review the usage for free to advise you on the best course of action (if any).

Setting Thresholds

Unfortunately, the defaults out of the box for the Advanced Monitoring will generate a number of false positives. All systems managed by Conetix should have these defaults updated, however if you have external systems then you can also update with these recommendations:

The key changes here to note are:

ThresholdOriginalAdjusted
Apache and PHP-FPM Memory Usage20%60%
MySQL Memory Usage20%40%
Swap Usage80%Disable

On a Windows Server, the ‘Apache & PHP-FPM Memory Usage’ will instead be replaced with ‘IIS Memory Usage’ – simply set this to the same values as for ‘Apache & PHP-FPM’.
Additionally, there will not be an NGINX Memory Usage option.

These changes better reflect real world scenarios and prevent false positives from triggering alerts. These changes can be made using the following instructions:

  1. Login to Plesk as an Administrator.
  2. Select Advanced Monitoring from the Server Management section in the left hand menu:
  3. From the right-hand side, select Settings then select Thresholds:
  4. Add the values from above and hit apply.

Detecting Resource Changes

If you’ve updated the server with additional resources (eg additional disk space), these won’t automatically be reflected within the Advanced Monitoring tools. To see the new resources, you’ll need to do the following:

  1. Login to Plesk as an Administrator.
  2. From the left hand menu, select Extensions then My Extensions:
  3. Find Grafana then select Open:
  4. If the changes have been applied, you’ll the status as a red exclamation mark indicating a change. Click on Restart Integration to re-read the values:
  5. The changes should now be reflected within the graphs and alerts.

Disabling Email Notifications

As of the latest Advanced Monitoring 1.2.0 (released December 2019), email notifications have been enabled for the alerts. This means you’ll receive an email every-time a threshold has been hit. If you haven’t adjusted the thresholds or simply prefer not to receive the notifications via email, you can disable them using the instructions below:

  1. Login to Plesk with an Administrator account.
  2. On the left hand menu, go to Tools and Settings:
  3. Within the Plesk settings area, click on notifications:
  4. Scroll to the bottom of the list and de-select the two notifications for resources:

    Tip: The first column for the emails is the Plesk Administrator.
  5. The emails for Advanced Monitoring resource alerts should now stop.

Was this article helpful?

Related Articles