We’re experiencing issues regarding an increased loading time of the search page after upgrading from 3.1.3 via 3.2.6 and 3.3.13 to Graylog 4.0.7.
The other pages load fine and feels “snappy”, the issue seems to be related to the search page and the dashboard views.
The time before the search page is fully rendered is about ~10-15 seconds. I have not noticed any unusual increase of load on the servers and the underlying Elasticsearch queries takes only the expected milliseconds to finish.
I’ve noticed similar behaviour as described here Rebuild fields index / slow '/api/views/fields' requests and here Large number of streams causing major performance issues with fields request · Issue #8063 · Graylog2/graylog2-server · GitHub regarding the API calls to
/api/views/fields and a large number of returned fields, almost 4000 in our case. Using the web developer tools in the browser I observed the page seemed “stuck” executing the API call to
I’ve tested the GUI via a load balancer (HAProxy) and directly to each node with the same result.
The environment consists of three physical servers with Graylog, MongoDB and Elastiserach installed on each server.
- CentOS 7.9 (SELinux disabled)
- Graylog 4.0.7
- MongoDB v3.6.23
- Elasticsearch 6.8.16
Any ideas on how to fix or troubleshoot the above mentioned issues regarding slowness in the GUI?