Select Language:
If you’re managing an Aurora MySQL database, it’s important to monitor the AuroraMemoryHealthState metric. This value tells you about your database’s memory health, with numbers ranging from 0 to 10.
A reading of 0 means everything is normal—your database has enough memory and is working smoothly. A value of 10 indicates a RESERVED state, which signals that your server is running low on memory and nearing a critical point.
Values between 0 and 10 show different levels of memory pressure. As the numbers increase, it means the system is feeling more strain. These intermediate values suggest your database is under some memory stress but hasn’t yet reached the critical level.
If you’re seeing your memory health state pinging between set points like 5, bouncing down to 0, then moving up to 10, this shows your database’s memory usage fluctuates over time. The pattern (such as 0 → 1 → 0 → 2 → 0 → 3 → 2 → 0 → 5) reflects the changing workload and how it affects memory availability.
This metric is tied closely to Aurora’s Out of Memory (OOM) Avoidance feature. When memory use gets high and the AuroraMemoryHealthState number goes up, this feature detects the situation and can take actions based on your settings in the aurora_oom_response parameter. These actions might include logging memory-heavy activities, adjusting cache settings, refusing new queries, or terminating ongoing ones to free up memory.
In newer versions, specifically Aurora MySQL 3.06.0 and above, there’s also a status variable called aurora_oom_status. It provides a quick overview with states like NORMAL or RESERVED, helping you easily understand your memory status at a glance.
Regularly checking this metric allows you to spot early signs of memory issues and act before your database runs into more serious problems. Staying on top of memory health helps keep your database running smoothly and efficiently.
ChatGPT
Perplexity
Gemini AI
Grok AI
