There are plenty of monitoring tools out there, some better than others. But the initial challenge is figuring out which ones work for certain kinds of tasks. There is a lot of functional overlap, and you don't want to spend time doing the same task twice. Of all the debugging and Python monitoring tools, you'll likely end up needing just a few, not dozens. The answer lies in categorizing the tools. After that, we can examine the different jobs that each kind of tool is designed for.
To get the most out of Python monitoring, consider the four main types of tools you can use to get the job done: tools that monitor metrics, uptime, tracing and logging. Logging tools tend to come in two flavors. Some are generic and others are specifically designed to detect errors. Tools for monitoring metrics are rather straightforward and include about a half-dozen very popular choices. For tracing, you can opt for so-called pure tools or the APM variety. Again, the marketplace offers multiple choices in this category. For uptime monitoring, there are hosted, self-hosted and open-source versions available.
Metrics
In essence, metrics are nothing more than carefully aggregated events from the log. If you're using a Python application, you have all sorts of choices for metric storage, including both hosted and open-source options.
Uptime
Even hosted solutions for uptime monitoring are surprisingly inexpensive. However, a one-minute online search will yield many open-source and home-made versions you can download for no charge to get started. Whether your site is massive or tiny, you will need an uptime monitoring tool. In fact, if you only use two different kinds of tools, your arsenal should include one for log monitoring and one for uptime monitoring.
Tracing
It's rather simple to generate metrics from aggregated transaction traces. In fact, when taken as a whole, this entire group of metrics is the backbone of APM (application performance monitoring) tools. There are two basic types of tracing tools. One simply displays, after collecting, various transaction traces. The other is a bit more complex because it shows general app metrics along with the individual traces.
There is currently not an accepted standard when it comes to transaction tracing. If you're used to dealing with metrics and logging chores, you're probably used to so-called industry standards. But that's clearly not the case with tracing tools, which makes shopping for an individualized solution something of a challenge.
Logging
Last, but neither least nor uncommon, is logging. It's safe to say that logging is the most common of all monitoring tasks for Python applications. For the most efficient log monitoring, there's good sense in aggregation. Only after aggregating your logs will you be able to efficiently perform filtering operations.
If you are currently shopping for a log aggregation solution, you have a lot of options, with self-hosting and open-source being a priority. The open-source versions are available in error-monitoring or generic versions. If your preference is hosted log aggregation, it's the same situation: you can pick from either generic or error-monitoring products. However, there tend to be many more hosted products available.
