Normal RAM Usage

Hello, I have Discourse installed on DigitalOcean. I see the minimum requirements are 1 GB of RAM, although it seems my server is using more than expected. Any thoughts, is this normal? Thanks!

What about that screenshot do you consider to be abnormal?

2 Likes

Hi Matt, I don’t know if you had a chance to read the rest of what I wrote, but I read somewhere Discourse only requires 1 GB of RAM. If that’s the case, it seems my installation is using more than expected. I don’t have any users yet. Before restarting the server, some of the swap memory was being used as well. Perhaps I didn’t know Discourse needed so much memory, and that was my question: are these normal results? Most of the allotted memory is being used. Thanks!

Check the existing topic below

Yes, that’s normal.

What you can do is tune down the unicorn workers number. In one instance I have:

  • 280MB per unicorn worker

  • 200MB sidekiq

  • 180MB unicorn master

I guess going from 4 workers to 2 can shave you some memory.

PS: I’m assuming that you are running a ghost town, when you get more users revisit this.

3 Likes

Thank you. Your response has been the most helpful. Any advice on getting it working on Linode – maybe even how to migrate the forum? Linode uses a different kernel, so it’s going to be a manual installation, and the kernel has to be switched as well. I can get double the memory with less storage on Linode instead of DigitalOcean, and the performance is also better.

1 Like

I have followed this guide soem months ago, and it worked fine:

However, keep in mind that people have some concerns with Linode security:


About migration, it’s easy. After having Discourse running in both servers, backup from old, restore on new.

3 Likes

I saw that tutorial, but it’s hard for me to use something where the person who wrote it doesn’t understand what’s going on. Here’s another resource I might try. Thanks for the help!

In regards to the security issue, I think this article on Linode’s blog may have a few answers.

1 Like