NxCloud dashboard slow or not loading but only using a single CPU

[Complex-Nerd]

Hi All

We are running NxCloud in a cluster with approx 3000 users and for as long as I can remember the dashboard and and logging screens take forever to load, all the other pages are fine. I am assuming it's because it it trying to interact with the db to produce the reports, which makes sense. The issue we see though is that it appears java is the main process and the system only really uses a single a CPU (out of a possible 6 available). Our monitoring reports that the single CPU hits 100% whilst the other 5 are just ticking over with minimal use.

Is there anything that can be done / checked to see what the issue is or why the app only seems to use a single CPU?

Thanks

Noel

[support200]

It's between JVM and your system. I don't know if we can force it. This guy is talking about some possible solution, https://www.databasesandlife.com/jav...ecs-multicore/

You can set those variables in /nxcloud/bin/startup.sh

And what's your NxCloud version? We separated operator traffic db with the latest version. Each operator has their own traffic db space. However, this doesn't solve your problem.

The best way would be separating logging/report using Graylog like tool. https://nxfilter.org/tutorial/h-gray...te-logging.php

If you can do it for operator GUI, that's even better. Or you can use Graylog instead of your admin GUI and let your operators use NxCloud GUI. In that case, you can bypass dashboard.jsp on /nxcloud/webapps/admin.jsp. There's redirection code.

if(actionFlag.equals("login")){
  if(dao.login(paramString("uname"), paramString("passwd"))){

    if(isAdmin()){
      // Start page for admin.
      response.sendRedirect("dashboard.jsp");
    }
    else{
      // Start page for operator.
      response.sendRedirect("zop,dashboard.jsp");
    }

    return;
  }
}

You can forward it to operator,operator.jsp. Then you bypass that report generation.

[support200]

Did you increase memory allocation? Just yesterday I looked into some company's NxFilter. They do cloud filtering business with NxFilter. They also have more than 3000 users. But I don't see any slowness. They allocated 4G of memory.

[Complex-Nerd]

Working on the Graylog integration has probably not helped the CPU and the reason I spotted it.

I configured NxCloud to log to Graylog and I was super impressed and my aim was to run both logging in NxCloud and Graylog untill i had done the work creting my dashboards in Graylog for the different Operators, leave it a while and then perhaps set retenion on NxCloud to 0 and just run log in Graylog. The dashboard was slow in any event but when I stopped and restarted the NxCloud, the dashboard just became unresponsive. I left it to settle for a long while and just watched a single CPU just max out. In the end I had to rollback my VM to get the dashboard and the logging in the NxCloud GUI back (the slave also couldn't communicate with the master but again put that down to CPU usage perhaps?)

We are running a version before the feature of seperated operator traffic. An upgrade was my next job but possibly this will need bringing forward. Are you saying if I bypass the dashboard with the above script I wouldn't lose logs in the operator GUI? (assuming I leave 10 days of retention in the config)

[Complex-Nerd]

Sorry, just read your script, I see that this is a redirect for the admin login so would help the admin. Dashboard and logging in the GUI for operators is also slow, though not as slow as he admin.

Were running v4.3.3.9 if that helps

[support200]

Yeah, you will not lose the data. If you bypass dashboard, it doesn't need to create charts for dashboards. Try that way first.

[support200]

Before you update it, read this. https://nxfilter.org/forum/announcem...ud-to-v4-5-4-2

I guess it will be faster for operators. They don't need to search their data from main traffic db. They have their own db space for traffic. Should be faster.