smp
commented
3 years ago Hi @HaterMonestary we really appreciate you taking the time to provide your feedback. Can you help us identify any reproductions of the issues you call out specifically? It looks like there are two main issues we need to track down along with a handful of documentation improvements that could help future users:
- Resource recreation errors does not seem to be caught or checked to see if the resource already exists (feel free to use this ticket to expand on issue)
- Buckets missing their bucket events to trigger the lambda to transcode the video (please create new ticket with any details you can provide on how/when/what's happening)
- Document how Amplify Video manages underlying resources to provide a video stream (some of the basics are here would love it if you could give us more specifics on the questions you have regarding the project in a new ticket that we could address)
- Document common customization approaches #121 would address templates specifically, and we have some bits on audio streaming in the wiki. What other types of customization are you thinking about?
- Document production vs non-production changes and best practices for going into production (currently folded into the wiki pages, but could be more explicit)
THOM-AwS
gvasquez95
My feature request is related to stack redeployment problems. I have been having issues with my stack updates, things are attempted to be recreated where they already exist, (my buckets, this caused me a lot of lost time trying to work out why my stack is trying to recreate or overwrite a resource that does not allow that),
I have noticed a lot of issues with amplify in general, but specifically with the vod sample I have followed I have hit a lot of potholes. this main one for resource recreation errors does not seem to be caught or checked to see if the resource already exists. also, I have for some reason issues with the buckets missing their bucket events to trigger the lambda to transcode the video into the output bucket. they just went missing and it would not start a transcode job. I had to copy them in from my dev branch and recreate them by hand to get the system to run the transcode service. Can you add in some resource checks to see if the required resource attributes are still available, and recreate conditionally if they are missing. There is limited documentation on customising the vod project unicornflix. having to go spend weeks learning how the interfaces between each part of the project has become pretty cumbersome considering how easy this was made out to be. it feels like sometimes any amount of deviation from the path of the sample code leads to errors that are hard to trace out.
The only other option here seems to be to just recreate them by hand, but you have to know exactly how it was before to get it to work. that is a real problem.
the unicorn flix project is mostly well done, but any deviation from the tutorial and you can get lost right away. The documentation is factually good, but offers very little in terms of how it works in terms of what does what, how with what, and how to customise it for your own purpose. There is virtually no walk through for production deployments, and what the feature differences are for prod vs anything else when it asks you if it is a prod environment. What does this mean for me?
I had to dig through a lot of documentation to make headway to bring this anywhere near being prod ready, and some pointers or walkthroughs on how to use the project with R53 and cloudfront would be helpful. there is virtually no info I could find, on how to deal with the cors issues I am up against now, for which I have tickets in with support to try to get it working. Some more detailed docs on how to bring this into prod would be really really useful.
I understand that some of this wont affect the amplify-video team, but I am trying to be a bit holistic with my experience in the hope that will be helpful for the next guy.