Dashboard and Mongo memory usage


We have standalone Tyk v1.9.1.1 with dashboard v0.9.7.2 using mongo v3.0.7 for storage and also a portal page and documentation site set up. Currently the size of the analytics data is a bit over the amount of available memory in the machine. If we access the dashboard mongo starts caching roughly 95% of available memory. The actual memory used never goes above 50%.

The problem here is that the dashboard loads for a long time and then doesn’t show any data. In addition, the “Users”, “Apis” and other pages show an error that none (users/apis) were found. Also the dev portal and documentation site stop working. The logging to file doesn’t seem to be working on Debian but the service log showed an error: “Failed to retrieve object: read tcp> i/o timeout”
A while back we deleted some old data and it started working properly but soon started caching most of the memory again.

What could be the problem here?
Does the dashboard need all analytics data to be in memory cache even though by default it only shows past month’s data?
Should not being able to load analytics data break the other dashboard pages, dev portal and documentation site?

Thanks in advance.


Hi Tarmo

Later versions of Tyk include analytics aggregation and report generation features that resolve this issue at scale.
The simplest solution would be to upgrade to the latest 2.x version.