It seems that the Docker was not able to execute all parallel builds.
The solution I found is:
Run the setup with option: ./nodebb setup --skip-builds Manually run the build in serial mode: ./nodebb build --seriesThat solved the issue.
I run NGINX with PageSpeed and HHVM.
My system has 4GB RAM, 2 vCores (Intel Xeon E5-2690v2) and a 25GB RAID 10 SSD. I can handle with this setup easily 1.000.000 users a month, traffic was optimized (I now have a CDN) and so stay around 250GB bandwith for my website only.
I rather not add the firmware archive here...
It consumes more than 10TB traffic a month.
Before I forget it, usually 100-500 people are online. And I never experienced a resource shortage, actually it is often the case that the RAM is only used to 25% and the CPU to 20%.
@AOKP said:
I run NGINX with PageSpeed and HHVM.
My system has 4GB RAM, 2 vCores (Intel Xeon E5-2690v2) and a 25GB RAID 10 SSD. I can handle with this setup easily 1.000.000 users a month, traffic was optimized (I now have a CDN) and so stay around 250GB bandwith for my website only.I rather not add the firmware archive here...
It consumes more than 10TB traffic a month.Before I forget it, usually 100-500 people are online. And I never experienced a resource shortage, actually it is often the case that the RAM is only used to 25% and the CPU to 20%.
Redis or mongo?
@exodo said:
@AOKP said:
I run NGINX with PageSpeed and HHVM.
My system has 4GB RAM, 2 vCores (Intel Xeon E5-2690v2) and a 25GB RAID 10 SSD. I can handle with this setup easily 1.000.000 users a month, traffic was optimized (I now have a CDN) and so stay around 250GB bandwith for my website only.I rather not add the firmware archive here...
It consumes more than 10TB traffic a month.Before I forget it, usually 100-500 people are online. And I never experienced a resource shortage, actually it is often the case that the RAM is only used to 25% and the CPU to 20%.
Redis or mongo?
Redis. I know that in-time memory is a little problem, but my systems never encountered any issues and that since more than 1 year.
A general performance tipp:
Never ever use CloudFlare, except you want to attrack spambots and lose performance.
For me it resulted in spambots visiting my site and so eating resources and yes, the result was the same for me on several servers, setups and websites.
Nowdays I use no additional spam protection like CloudFlare and I never encountered any issues within the last 1 and a half years (even in myBB).
Last time I tried CloudFlare was 3 months ago, just to have it said.
Btw how HHVM benefits you on nodebb? Isnt it for php?
@exodo said:
Btw how HHVM benefits you on nodebb? Isnt it for php?
Of course it is only for PHP. However, in my case PHP Scripts as the shopping system (WordPress + WooCommerce) are also a part of my site and therefore should not be forgotten when talking about traffic/load.
On a $5 DigitalOcean droplet, we've seen the forum handle loads up to 300-400 concurrent users just fine, and this is on a single CPU!!
Any more and you should probably get more cores, move your db to its own server, or scale horizontally