
ballchasing.com
@BallchasingRL
Followers
913
Following
19
Media
15
Statuses
65
site: https://t.co/K5eq0HXswz discord: https://t.co/3mxnk9PHWB patreon: https://t.co/u4YrYJT0nn
Joined January 2020
I used to use and pay for DataDog to collect metrics and monitor the site, but that got too expensive then switched to NewRelic, which also got too expensive with time. I'm now using the fee plan of NetData
0
0
23
then there are a 2 worker servers handling the replays parsing, indexing, stats computations, ... 2 front servers serving the website and api, behind a load balancer all of these are hosted using Hetzner too The Hetzner stack costs around 400$/month Continued ...
2
0
21
The replays listing/searching uses an Elasticsearch cluster running on a 3 bare-metal servers, from Hetzner too. Continued ...
1
0
19
Once a week, an automated job (Github Action) is triggered which ensures that the database backup can be restored to a new and freshly created server, which is then discarded at the end of the test. Continued.
1
0
20
the database is postgres, running on a bare-metal server from Hetzner and weighs around ~575GB as of today the database is backed up to Backblaze object storage (used to be in Wasabi too, but I'm planning on ditching the latter due to their 3-month deletion policy Continued ...
1
0
21
The replays + generated files (heatmaps, 3d replay viewer data, ...) are stored in wasabi, an S3 compatible object storage without egress charges Still, this is the most expensive part costing ~800$/month Continued ...
2
0
35
In this tweet, I'd like to publicly answer a question I often receive in private. Q: How/Where is https://t.co/igFk79RaCc hosted ? Cloud vs Self-hosted, the cost breakdown, what technologies are used, ... Check the thread for the answers:
ballchasing.com
Rocket League® replay statistics (Boost, Positioning, Ball, Demos, Settings, ...)
6
10
205
I don't like it but I had to resort to a huge banner in the website asking for patreon support because keeping the site up has been taking a toll on my finances for a couple years now. I've been consistently paying a couple hundred euros from my pocket every month for a while😓
41
230
1K
Lessons learned: - disable the automated terraform apply to start - will look for more ways to avoid future accidental server deletions
0
0
5
- the automated apply noticied that the OS should change from ubuntu 20.04 to 22.04 - so it went ahead and recreated all of the frontend (website + api) and worker servers - hence the downtime
1
0
3
The issue: - I changed & pushed the default OS to ubuntu 22.04 in the terraform config - I forget to explicitly set this variable value to the current 20.04 for the prod env - An automated terraform apply was executed yesterday at 7pm - ...
1
0
4
After 5 hours & 30 minutes of downtime, the service is up again. What happened is an unlucky sequence of events. Check the thread for more details. My apologies for such an extended downtime.
4
1
43
Alright let's try to steer conversation back to the RLCS for a bit: What's your favorite "advanced stat?" @BallchasingRL keeps track of player speed, distance to ball, time in different 3rds of the field, boost averages, all sorts of fun stuff. What's your favorite?
18
1
120
❤️
When finding footage for our videos, my job is made infinitely easier when I utilise features from @LiquipediaRL, @octane_gg & @BallchasingRL LQ for scores, Octane for specifics, and Ballchasing to check the 3D view & to download the replays from. They're all invaluable 🐐🐐🐐
0
0
2
❤️
Many thanks to @BallchasingRL for the amazing work through the years, from building the website as the Rocket League replays hub to designing the associated API.
0
0
4
Nearly 8,000 Replay Files! Get access to the #RLCS 2021-22 Replay Group on @BallchasingRL! https://t.co/bHU3IprUey
0
1
23
Brace yourselves for the Dark Mode update is dropping !
3
3
58
Damn but while working on the upcoming dark mode, switching back to the light mode (to check for regressions) hurts the eyes something fierce !
1
0
18
A dark mode is in the works. No ETA yet but it is coming *soonish*.
3
2
109