If you want to avoid problems and keep your systems up and running, monitoring is a must. While there are many different things that you can monitor, here are five that we have found to be particularly vital for Atlassian’s Jira and Confluence:
- HTTP Availability – Jira and Confluence are both web-based applications. You therefore need to monitor that the site is up and available, the page is loading quickly and the login authentication is working properly.
- Database Size – There’s a setting on the SQL database log files called “MAXSIZE.” This is usually set during the initial application installation. We’ve run into issues where Jira hit this size limit and went offline. Users couldn’t log into Jira because the application could not write any more transactions. To fix the problem we had to go in and make a modification to allow the log file to grow bigger.To avoid this problem, set your monitoring system to alert you if the database reaches 80% of its maximum size. It’s also a good idea to have a backup in place that will regularly free up log space.
- Disk Space – Disk space is a vital thing to monitor, because if you run out of disk space your services will stop working. While the disk space issue affects both your transaction file database and your log files, it can be particularly problematic in the folder where the logs are generated. In fact, we’ve seen it happen where an organization runs out of log space, Jira or Confluence goes down, and someone had to clear out the logs in order to get things up and running again. Just like with the database size, our recommendation is that you set your monitoring threshold for 80% usage. Because things can go wrong in this area very quickly, this warning threshold gives you a chance to react if disk space starts to suddenly and rapidly disappear.
- Disk I/O – There are two things that you want to look at in the disk I/O area. First is the amount of data being transferred back and forth. The monitoring goal is to ensure that the I/O is kept below a certain threshold. If it is constantly high, this is an indication that there is either congestion on the disk subsystem or that the system is being continuously taxed. Neither is a good thing!The other thing to monitor here is what’s called “latency,” which is the time (in milliseconds) that it takes to complete a single I/O operation. High latency numbers indicate that it’s taking the application longer to access data on the disk, which will result in slow performance for your end users.
- CPU & Memory Usage – With Jira and Confluence, you monitor CPU and memory usage by monitoring the JMX (Java Management Extensions). Jira and Confluence both use the Java engine to run most of the website. Monitoring the JMX will tell you how much CPU and memory your Jira or Confluence instance is using.The goal is to ensure that CPU and memory usage are not too high. A safe, “normal” level is between 40 and 70 percent. While brief spikes in usage are okay, you never want it to be continuously spiking to 100%. Continuous spiking will deteriorate the performance of the Jira or Confluence website itself.
Need help getting all of this monitoring set up – or want to outsource all of the actual monitoring once the setup is complete? Give us a call. This is what our Cloud Services offering is all about!