In the maintenance phase, active development of your application has ended, and the focus is on "Keeping The Lights On" (KTLO): ensuring your application remains available for users.
At this point in the life cycle, you should understand how your application behaves. You will need to decide what is considered an acceptable level of performance so you know when it is worth investigating an issue.
An acceptable error rate is the percentage of errors deemed acceptable for an application. This rate varies per application and takes into account the context and consequences of the errors. You can apply these rates to your entire application or have different rates for different application components.
To contextualize this: an e-commerce application may have an acceptable error rate that's as high as 10%, whereas an application that processes medical data may have a substantially low rate, close to or equal to zero.
To determine what constitutes an acceptable error rate, you need to thoroughly understand your application's architecture and the consequence of any errors for your end-users. Weigh this up against your technical constraints, such as the limitations of your codebase and the practical and financial costs of maintaining a low error rate.
AppSignal automatically tracks your application's errors, allowing you to see your error rate in real time. You can also specify if an error is critical or something that can safely be ignored by giving errors one of the following severity labels:
In the maintenance stage, you should monitor your application's errors to ensure the rate remains within your definition of acceptable and critical functionality remains available.
An increasing error rate on an application that is no longer being actively developed could indicate issues with your stack, unexpected changes in usage, or memory problems.
You can read more about error rates in our article: What are good and acceptable error rates?
Your host metrics are important, and while your application is in maintenance mode, your hosting requirements may change. To keep on top of changes in demand and make sure you aren't using more hosting than needed (or slowly running out of disk space), consider keeping an eye on the following host metrics:
|Metric||Description||What to look out for|
|Disk usage||Monitor your disk size. Your application may require more or less disk space.||Watch for sudden increases or decreases in disk usage, which may indicate potential issues. For example, a significant increase in disk usage might suggest logs or temporary files are not being cleaned up properly by your application's garbage collector.|
|Load average||Load average tells you how many tasks are waiting to be processed or being processed by your application. Monitoring your application's load average can help you ensure that the application is performing optimally and can handle the current workload.||High load averages may cause performance issues, such as response delays or system crashes. Monitor for load averages exceeding your system's capacity or showing a consistent upward trend.|
|Memory usage||Monitoring your memory usage during the maintenance stage of your application helps you prevent your application from using too much memory, which can cause the application to become slow or unresponsive, negatively impacting end users.||Excessive memory usage, like continuous growth or sudden spikes, can lead to performance problems such as slowdowns or unresponsiveness. To address this, take steps to optimize memory usage, detect and resolve memory leaks, and streamline data storage.|
This article is part of a series. We have a Learning Center article for each stage in the product life cycle:
Join us with a free trial to start monitoring your application with AppSignal. You can be set up in as little as five minutes. If you have any questions or require help, our dev-to-dev support is on hand.
Do you need help, have a feature request or just need someone to rubber duck with? Get in touch with one of our engineers.