Increased CPU usage on multiple instances and accounts
Hi
Thanks for reading this.
Last monday, I notice that all of my RDS databases have an increase CPU usage, is something low but it's there. The increase started January 27 at the same hour (08:00 UTC) in 6 diferent accounts. The rest of the metrics in cloudwatch don't show any changes.
I have restarted and apply some pending patches in a few of the DB to see if that fix the issue but with no luck. The issue happens across diferent engine/versions: MariaDB (10.6.22; 10.5.29; 11.4.5), PostgreSQL (15.15) and diferent instance types (all in the t category): db.t3.medium, db.t3.small, db.t3.micro, db.t4g.small, db.t4g.micro.
Yesterday I created a new DB (this don't have any data on it and is not use by any application), the CPU usage is similar to other that is used by a web site.
Someone has any idea why this happens?
- Newest
- Most votes
- Most comments
Based on the information provided, there are several potential causes for the simultaneous CPU usage increase across your RDS instances:
Common Causes of CPU Spikes:
The most likely causes of sudden CPU spikes include:
- Connection storms - Your applications may have opened too many simultaneous connections to the databases
- Workload changes - New queries, increases in dataset size, index maintenance or creation, new functions or operators, or increased parallel query execution
- Query execution plan changes - Changes in how queries are executed (for example, switching from index scans to sequential scans) can increase CPU requirements even for the same operations
Recommended Investigation Steps:
- Review key metrics - Check for spikes in:
- Queries
- Com_select (for MySQL/MariaDB)
- Innodb_rows_read (for MySQL/MariaDB)
- Number of database connections
- Investigate application changes - Since the increase occurred simultaneously across multiple accounts at the same time (08:00 UTC on January 27), examine whether:
- Any applications connecting to these databases were updated or changed behavior at that time
- There were any infrastructure or network changes
- Connection pooling settings were modified
- Check for backend processes - Determine if the number of backend processes or connections has increased, which could contribute to sustained higher CPU usage
Since you've observed this behavior even on a newly created database with no data or application connections, and the pattern is consistent across different engine types and versions, this suggests the issue may be related to a common factor affecting all instances rather than database-specific workload changes. Consider reviewing any centralized monitoring, backup, or management tools that might have changed behavior across all your accounts at that specific time.
Sources
Resolve high CPU usage after Amazon RDS for MySQL and Aurura MySQL-Compatible upgrade | AWS re:Post
CPU - Amazon Relational Database Service
answered 8 months ago
Relevant content
asked 3 years ago
asked 2 years ago
asked 3 years ago
- AWS OFFICIALUpdated a year ago
I don't see an external cause that explains the increased CPU usage. What other explanation could there be?
Thank you for any help.