• Off-topic
  • Which OS is best for flarum working ?

My Setup

Operating System: CentOS VPS
CPU: 6 Core
Amount of RAM: 12GB
Connection Speed: 1GB/s

BTW which OS is best for Flarum ?

Linux has many distros, please specify which one. Ubuntu? Red Hat? CentOS?

Distros? Anything that runs containers.

300K posts forum on i7-8700 CPU @ 3.20GHz, docker containers in LXC container.

    Prcek I have 6 Core CPU and 12GB RAM ... still mine VPS is very very slowwww

    Which VPS do you use ... Please tell.

    I have 100K registered users and around 5k memebers or users are online per day

      xasharma It would be worth checking to see if any extensions are slowing down your instance.

      For our instance I didn't want managed VPS, but a proper bare metal and Proxmox Virtual Environment on top of it. Hetzner dedicated server comes out pretty much the best. That's why it was chosen. Default Flarum setup, no caching hacks, everything runs smoothly.

        Prcek dedicated machine is far too overpowered for Flarum. And I don't think containers is something to be mentioned in a discussion about OS or server requirements. Containers is something for a professional and/or highly scalable environments with thousands of concurrent users. Using these terms in this discussion only derails the topic and confuses people.

        As said a few times before, I think your machine @xasharma should be more than enough, something else is in the way. Maybe MySQL or a proxy. Try running Flarum on a digital ocean droplet with 2 CPU and 2 GB RAM to see if that helps; you can cancel it anytime.

        Flarum's hardware/server requirements are really low, keep that in mind.

          xasharma well I know your website its down rn may ik what you use on the vps to run your website is cpanel cloudpanel what is it?

            luceos Flarum's hardware/server requirements are really low, keep that in mind.

            With all due respect, I can't agree. If a forum has dozens of active users at the same time, the API response time increases significantly. A powerful CPU and enough threads solves this. This is the experience of two years of operation.

            The matter of containers is a matter of professional approach. The sooner the administrator gets to this point, the more free time and peace of mind they will have 🧘🏼

              Prcek If a forum has dozens of active users at the same time, the API response time increases significantly.

              Could this be because of synchronous driver? It’s probably not power and threads that are needed but just a queue (hence asynchronous operation).

                Prcek Personally, I think it doesn't make much sense unless it's a use case of multiple applications hosted on a single server (where it starts to become a headache to manage all the different services...), or already other obvious/necessary cases where it's not clear how much the service/application can grow. It reminds me a bit of people using Kubernetes for super simple applications, it's hard to believe that there are people willing to go through that torture voluntarily.

                For the 1 server 1 application use case, I don't think Docker is a big deal, apart from losing a bit of server performance, running a single application on a server, especially Flarum, should be pretty straightforward and predictable, I mean by consuming so few resources, I think you'll be able to see well in advance when you need to upgrade or even use flexible or burstable servers for unexpected peaks.

                  CyberGene That's legit. I'll dive in. Thank you.

                  Darkle I sympathize with any application running on Kubernetes because once an administrator gets used to a declarative configuration, it's hard to go back.

                  luceos something else is in the way.

                  Thats i don't know ... how can i get to know whats the problem ?

                  xml cpanel cloudpanel what is it?

                  Its cpanel

                  Prcek With all due respect, I can't agree. If a forum has dozens of active users at the same time, the API response time increases significantly. A powerful CPU and enough threads solves this. This is the experience of two years of operation.

                  Sorry wrote wrong there xasharma ... The number for online users is Around 5k per day.
                  And all the notification for all things are stopped on mails. Its only on site dashboard.

                  • HD3D replied to this.

                    CyberGene Could this be because of synchronous driver? It’s probably not power and threads that are needed but just a queue (hence asynchronous operation).

                    I turned on Redis queue/cache/sessions and 2x workers. API read response times did not change. On writes the times probably got faster, but that was never critical. So from my point of view, it's really true that a big hammer (CPU) is right for a big forum.

                      HD3D Well I've thought about lscache. I'll get on it. Actually, this is the last bigger issue related to Flarum operation which I haven't seriously dealt with yet 👷🏻‍♂️ Thank you for the push, sir.

                      HD3D Previous i have on a OpenLiteSpeed server, with 1GB RAM and 1 Core CPU but it was there slow also

                      Prcek just to add some background information to this statement.

                      Flarum, for the most part, is very performant. There might be some edge cases where perfomance might still be impacted, but once reported we always do our best to patch issues as soon as we can.

                      Aside from that, from my experience, it's usually community extensions (or the amount or a certain combination) that is detrimental to the performance. In our hosted flarum stack (on azure Kubernetes) we monitor these outliers, report and usually contribute patches. Regardless, this is an ongoing activity considering new and updated extensions.

                      Perfomance will always be a key area for core and your help in identifying extensions that can be optimized is crucial in guaranteeing a healthy Flarum ecosystem.

                        luceos Thank you for your contribution to the topic. I have reason to believe that the Flarum code is clean and free of major technical debts. If I was talking about performance, I was referring to API response times (Flarum backend + PHP + MySQL), not Flarum itself. The developers deserve one big thank you and more financial support from Flarum users.

                        I jumped into Flarum with a forum migrated from phpBB and some 200K posts and 35K unique visitors per month sometime around Flarum beta 11. It didn't work well on weak hardware then or now. For PHP's sake and for MySQL's sake.

                        But if there were major performace flaws, it was always related to community extensions. I agree with that, of course. Perhaps it's just a shame that we can't centrally push events from Sentry and Clockwork reports so that developers can access them

                        Overall, I'm very happy with Flarum and I highly commend the work of the core and community developers.