My RDS(Aurora PostgreSQL) serverless reader instance database capacity count spikes every hour.
I have Aurora PostgreSQL reader and writer instance connected to node js application which uses TypeORM for db connectivity.
A few days ago, my serverless reader instance started spiking every hour(serverless database capacity) which I have set to 5-36 i.e. a minimum of 5 when there is no load and a maximum of 36.
I checked in performance metrics as well, there are no such SQL calls which is causing this. there are other times when there is more load but it does not spike, it always spikes every hour like 5:00, 6:00, 7:00,! Also, I have no cron job scheduled.
PS - The Writer instance works totally fine i.e. there is no spike.
Adding Image URL for same below.
- 新しい順
- 投票が多い順
- コメントが多い順
Kindly perform below checks and let me know your findings
- Check Scheduled Tasks
- Examine Connections
- Analyze the query running at the time of the spikes
- Review Auto Scaling Policy
- Check if any external services or utility is connected to the database
- Check Capacity Settings
If something is happening that is not tied in to SQL processing, that could be some memory or CPU usage happening in the underlying operating system. I suggest turning on Enhanced Monitoring (EM) and looking at what's happening at the operating system level.
I have seen where SQL submitted through Data API didn't show up in Performance Insights. So check if there any Lambda functions running that might be submitting requests through Data API. I like to use CloudTrail and filter by specific time windows to find things like Lambda functions I forgot were still running. Also check if there are any pg_cron scheduled jobs on the Postgres server, in addition to regular cron jobs on the client systems.
関連するコンテンツ
AWS公式更新しました 1年前
Yes, I have checked all of the points mentioned. There is no such anomaly or pattern.