Installation takes forever, then ends without forum actually starting

Installation
1

So I’m trying to start a forum on GCP’s Compute Engine, running it on an e2-micro. I think a 2GB swapfile was made when I first tried to run it.
10GB didn’t work because it was too little IIRC, so I increased it to 20GB.
After running ./launcher rebuild app, it started doing it’s thing. When it was doing ...[@embroider/webpack, it took so long I just left and come back later (~2h later).
I found that it was done, but I could not access my forum, even though I connected my Namecheap A record with the external IP of the VM.
I tried it again, and here are the full logs (logs are too long):
GCP e2-micro discourse logs.txt (176.7 KB)
And yet, my Discourse site isn’t working. I get:

502 Bad Gateway

nginx

1 Like
2

Now I get:

504 Gateway Time-out

nginx

3

I got it working! Never mind. I did nothing. It solved itself out.

1 Like
4

Does reducing Unicorn workers to 1 in containers/app.yml to conserve memory, and then rebuilding the app with ./launcher rebuild app works by any chance?

Or you can temporarily upgrade your instance to e2-small or higher, complete the rebuild, and then resize back to e2-micro.

2 Likes
5

Already rebuilding took ~2h. I don’t think I want to wait that long for the 3rd time.
On another note, my email now doesn’t work.

1 Like
6

Email works now, but I’m afraid of installing plugins, since rebuilding will take about 2 hours. I know this isn’t normal, so is there any way to speed it up without changing the resources of the VM?

7

Is this a fresh install? Or Fresh install with transfering the old site to the new server?

2hrs seems very long in my XP.

8

Fresh install. Maybe it’s because the server has 0.25 vCPU…?

2 Likes
9

That could be. If you don’t mind me asking what is the VPs system specs?

Quite awhile ago now. One of the servers I manage as a volunteer was on a basic 20g VPs. And even with how big the db was on there it typically only took usually around maybe 15m.

Eventually the client was forced to mitigate to a bigger server after it crashed hard. As they are the time didn’t heed my :warning: the server was going to crash as could no longer was able to rebuild via cmdline.

So they paid a member here to move it to a new server with 256g plan. Costed them a lot of downtime about 2.5 weeks for ignoring the issue.

On the plus for the most part they listen better now to :warning: when given.

10

I believe it’s 0.25 vCPU and 1 GB RAM. The setup made a 2 GB swapfile. The original server came with 10 GB storage. But the installation said it needed more to continue since there was like 1.7 GB left. So I increased it to 20 GB and I haven’t seen any storage-related warnings yet.