A lot of users have come across failed rebuilds recently (since September 2024) with the command was killed with SIGKILL error, even though the servers have more than 1 GB RAM (even 4 GB RAM).
Agreed. Though the setup for the VM i was using really should have some swap, the machine should have had more than enough RAM to handle the build. Maybe just call out the need for a swap partition/file in the docs?
I’m aware of this, but this doesn’t address the issue unfortunately. Instances with more than 2 GB RAM will experience the rebuild error, and even instances with < 2 GB RAM who already had a 2GB swap file created by discourse-setup experiences the issue as well, like @alltiagocom’s instance.
Plus, this problem seems quite recent and was never reported before September 2024, according to the search.
Yes, it does. But we are in a situation, where it should create at least 2 GB of cache every time there is less than 8 GB of RAM.
I would guess it comes from the fact that every single forum will use plugins. Or something, because I don’t understand mechanics of these things.
It should be made clearer that those values are real minimums in deed, and if an admin decides to use different plugins, more RAM and cache are needed. I mean, we have far too many topics where the reason for the issues is a lack of memory.
This seems to need to be increased. As @Canapin pointed out with my topic having 4g of ram was not sufficient.
And as I mentioned in my topic the instructions for adding a swap file should be added to git hub install instructions even if just an option if 4g of ram or less.
)