we have build a website to provide search in some specific area. For that we have used apache Solr (6.x), HAPROXY (1.6) and apache web server on a single system with CentOS as OS. our expected users are from 5000-10000 per day (at start). System is core-i5 with 12GB RAM. Out of 12GB, 8GB is given to Solr. HAProxy is configured with default setting and only some acls are changed to allow read operation from Solr only.
Currently two ports are open, one is for webserver (port 80) and one is for HAPROXY (port 9090). HAPROXY is used for Solr only that's why we have open a separate port for it. Live website configuration are in a virtual-host file in webserver folder.
Now we are facing performance problem to load the website and when user query for some search, it even takes a lot of time to establish a connection to out server. I have checked different timers in HAPROXY but total time it take for each request is very small (milliseconds) as we observed from user side.
Where is the problem? What are possible solutions to solve it ? I think this system (core-i5) is enough to server for these small number of users. Is there any configuration parameter in HAPROXY that is causing this problem ? Is my approach to use HAPROXY is correct?