Hello Reddthat! It is I, the person at the top of your feed.

First off. Welcome all new users! Thank you for signing up and joining our Reddthat community.

Bot Defence

Starting yesterday after going through 90+ registration applications, I couldn’t do it anymore. I felt compelled to give people a great experience with instantly getting let in and kept my phone on me for over 24 hours, checking every notification to see if that was another registration application I needed to quickly click accept on.

I want to quickly say thank you to the people who obviously read all the information and for those that didn’t I’m keeping a close eye on you… 😛
I found a better solution to our signup problems.

As we use cloudflare for our CDN I have turned on their security system for the signup page. ~~Now when anyone goes to the signup page, they will be given a challenge that needs to be solved.
That means any bots that cannot pass cloudflare’s automated challenge cannot signup.
A win until we get our captcha back working. ~~ Well I did not check the signup process correctly. It doesn’t act as I thought it would, so I’ll disable it after the migration.

Downtime / Migration to object storage

Today in the fediverse we have successfully confirmed that object storage will be an acceptable path forward but will not operate as initially hoped.
I initially hoped to offload everything via our CDN, but the data still needs to go through our app server. The silver lining is that we can still cache it heavily on our CDN to ensure that the pictures will be served fast as possible for you.

So it may be slightly pricier than we initially planned for when moving to object storage, but in the end we still benefit, functionally and monetarily. The reason is we were not going to be billed for egress (fetching/displaying images), where as now we will be. The fees are very low and still should be covered by our wonderful monthly donators.

We will have about 15-20GB of storage that needs to be moved and unfortunately our image service is incapable of running at the same time the migration is done, which means we need to turn it off while the migration happens. To top it all off we have… 43000+ (and counting) small image files. If you haven’t worked with large swarms of small images before, the one that I can tell you is that transferring small images, sucks.

So we can do two things:

1. Turn off everything

  • Dedicate all CPU and bandwidth to the migration
  • Ensuring continuity and reducing the risk of something going wrong

2. Turn off the picture service

We can run Reddthat without the picture service & uploads while we perform the migration, but the migration will have an impact in server performance.

  • This will amount to having any picture we host (that isn’t cached) return a 404.
  • Any uploads will timeout during that period, and return an error popup.
  • Pages will be slightly slower to respond.
  • Something else might break 🤷

Because of the risks associated with running only half our services, I’ve decided to continue with our planned downtime and go with option 1, turning off everything while we perform the migration.

Date: 28th June (UTC)

  • Start Time: 0:05
  • End Time: 6:00 (Expected)

It will probably take the whole 6 hours. In our testing, it did 150 items in 10 minutes… I will put up a maintenance page and will keep you all updated during that time frame especially if it is going to take longer, but unfortunately it will take however long it takes.

This will be the last announcement until we do the migration.
Cheers,
Tiff

PS. Like what we are doing? Become a contributor on our Open Collective to help finance these changes!