Question about monitoring mails and monitoring tools in webpage

Ops and Admin
1

HI!

We have a M30 Cluster in MongoDBAtlas, and from time to time we receive mails about a high CPU consume, like this (note the date and time):

mail
mail922×707 31.9 KB

But when we go to the monitoring tool, we don’t see for that date and time the high consume of CPU, instead shows for the normalized use of CPU only 4%.

monitortool
monitortool1630×700 73.6 KB

So, the first question is in which monitor system we have to believe or I’m missing something?

The second question is regarding these “empty” spaces in the monitoring tool, like in this image:

monitortooloverview
monitortooloverview1218×662 96.3 KB

As you can see, there are many days with low or nothing use, but our use of the DB is the same for all days. Then, why is no reflected in the monitoring tool?

And the third question is regarding the cluster sizes. If it possible to upgrade only CPU and not everything else?

Thanks in advance.
Regards,
Víctor Merino

2

Hi Victor, I recommend you to analyze the metrics for:

  • Opcounters
  • Query Targeting
  • Query Executor and
  • Operation Execution Times

along with CPU usage.

With them you can detect whether there has actually been an increase in queries and even identify problems with them. CPU bloating is a common symptom for inefficient indexes (of course, this is not a rule). From there, you can identify the periods and analyze the queries that took the longest through the Operations Log.

Regarding CPU-only upgrade, this is not possible on Atlas.

1 Like