Something Ocassionally Is Pegging the CPU Briefly - How To Find The Cause?
My Cpanel Server running MariaDB 10.6.18 (10.11 is still considered "experimental" on latest WHM Stable release) is having sporadic periods of CPU overload for 30 seconds to 2 minutes. Top monitor grows up near 100 when the average CPU load 99.95% of the time is about 0.8. Server is way underutilized (8GB RAM, 4 core VPS, Almalinux 8) serving an app for a small business with 20 users.
It get so overloaded I can't even SSH into the server. It doesn't happen at a consistent time of day that might correspond to a backup or cron event. It can go 1-7 days without it happening and could at worse happen 3 times on the same day. Happens during peak and during lowest user connections.
The VPS provider (which actually it's supposedly a VDS - Virtual Dedicated Server which I believe means dedicated CPU resourses) claims nothing is wrong with the VM itself or the hardware. Also have ruled out DDoS attack. Process list doesn't show any process pegging the CPU when it happens when I am able to see htop via console but it can come and go faster than I can get to the console sometimes but I did see the console when it was at 80 CPU (as opposed to the normal average of 0.8)
Is there anything other than the slow query log that I can look at to definitively rule out something going on with MariaDB. Memory use never goes above 75% total with Maria DB never using more than 4.6GB of the 8GB.
The pattern and behavior is looking more and more to me like an issue with a VM Noisy Neighbor or something haywire at the Hypervisor level. Only way to tell for sure is to change hosts. In over 15 years of using Cpanel I've never seen this behavior before.
-
Hey there! If you stop the MySQL service for a period of time, does that stop the issue? Knowing which service is causing the issue is likely the major point of the battle, as then you can go from there tracking down more specific causes.
0
Please sign in to leave a comment.
Comments
1 comment