An outage of some kind nuked my website and now I can't even find the custom WordPress I uploaded to the attached volume. As a result, I've had to refund a client their maintenance fee for the year. To say I'm annoyed is an understatement. Initially I was very impressed by the platform but now I will telling my colleagues and friends to avoid like the plague. Digital Ocean is harder to setup initially but 100% the way to go if you value reliability
AS
Alejo S.
Bad! Don't use it for personal projects
Bad! Not use it! Slow and Charge you for personal Projects
DC
Daniel C.
I have been using fly.io for almost 2…
I have been using fly.io for almost 2 years now, there have been ups and downs, its been a bumpy road. But this is expected for a new provider.
In the course of these 2 years I have deployed various prototype NodeJS applications.
I have been around the block with my deployment infra - never feeling fully satisifed with any of them, AWS Lamda, GCP KE, CF Pages, CF workers, Netlify, Vercel (best DX imo), Heroku, DigitalOcean
But time and time again I come back to fly.io
It just works, no frills. You write a dockerfile, you deploy it. and thats it. This is the serverless dream that I think most providers miss out on - they try to abstract away the infra too much, with Fly you get the best of serverless but still have full control over what you are deploying
Its incredibly fast. I cannot stress this enough. No cold starts, it kind of blows my mind.
The monitoring is simple. With CF Workers I can't count the number of times I will get the 500 Cloudflare worker page with absolutely no logs showing and I had to contact support.
I could write a lot more, but just felt compelled to leave my opinion on this great service.
Satisifed customer, I will be using it for all my clients.
If only they could have have a fully managed Postgress service I would be entirely fly.io stack
MA
Matt
Cheap, low latency, everything else is meh
I want to love fly.io.
We have a contender with a useful free tier again after Heroku (understandably I guess) got rid of it.
The http response latency is really, really good.
But the first day I hit problems with my servers not deploying that looking at their support site affected a lot of other users and ended up being platform issues.
Other customers are complaining that they changed defaults from dedicated to shared IP addresses, breaking their deployments and losing them business.
I'm unable to deploy my application again today (twice in 3 days!), with others in the forums failing to diagnose the same issue (the error messages are useless). One of them had "bricked their organisation" by deleting a builder while trying to fix it. At that point I gave up- I hadn't invested enough for it to be hassles moving.
I sincerely hope things improve and this is just a rocky moment in their growth, but I'd suggest checking the recent posts in their support forums to confirm things have changed.