Problems with memory leak in confluence. Constantly increasing resource consumption.

[Arthur Foss]

We have a server of confluence 6.4.2. After some time of work (about a month, maybe more), it starts to slow down, not respond when trying to open pages, logs contain OutOfMemoryErrors. After restart all is well.

We installed javamelody to collect statistics and saw that some systems and resources are constantly increasing.

Does everyone have this problem, or only some? What should I do to avoid restarting?

These some statistics that looks not normal in my opinion:

Swap space grows rather fast 

CPU load increases over time

Some other diagrams

Information about our confluence

Confluence Usage
Total Spaces	1433
Content (All Versions)	3496040
Content (Current Versions)	366249
Local Users	14720
Local Groups	1944

 

Add-ons (user-installed)

[kennyngkk]

Hi Artyom, 

Explanation 

After going through the information provided, I have a theory on what this happen

There are currently something in your environment is eating up all the ram/or you might not have enough memory, to begin with. From the look at the print screen above, it looks like Confluence is set with xms 1g and xmx to 8g. 

From the look at the swap memory chart. It looks like something is total wrong with your physical memory. For more information on swap memory, please have a look at the page below. 

• https://www.centos.org/docs/5/html/5.2/Deployment_Guide/s1-swap-what-is.html

From my past experience with the amount of space and users, 1g memory is not enough to support Confluence environment and this will cause constant garbage collection. 

Next Step

1. Make sure that your system has enough physical memory to support all the software
2. Change the setenv.sh Xms1g to Xms4g to ensure that Confluence has priority over the amount of ram when Confluence is the start of Confluence