API server doesn't start under docker-compose #156
Comments
|
Hi @deftdawg, I published new docker images. However, I never saw this kind of error when starting with |
|
I pulled the latest from the repo and reran The host OS is nixOS, a distribution of Linux. Here's what I have for my docker networks and interfaces: |
|
I actually found I still had the old images from 7 months ago cached, so I The new The really bad news is now both my NixOS and my Virtualbox docker hosts both see the The prior version of the webapp image from 7 months ago, used to start with only 1 line: If I manually start the container, I can get server.js to run if I use npm to install express: |
|
hitting http://localhost:3001 on my Virtualbox, I get bounced off to an OAuth2 server: to prevent myself from being logged into takeniftynote.net, I add an That gives me a I get logged in to the UI locally, some css and stuff are loading from it, but everything else is still going to Amazon in the EU (saveDocumentTitle -> 717zri8ug2.execute-api.eu-central-1.amazonaws.com)... The previous version of webapp seemed to do more locally (there were options for Ollama) even though the APIs were borked from my broken API server... 😢 |
|
Thanks for your effort and awesome catch! I added a tiny comment to your PR, if you would like to address I am happy to merge. Regarding the current images, I did a mistake with the environment of the image build. If you pull the webapp image again it should be fixed. Also the missing express is fixed. However the code change is not pushed, It can be done with your PR then 🙂 |
|
Good news is requests are correctly routing to localhost, bad news is everything going to 3002 is dying because of CORS 😢 API server shows this in the log, attempting to saveDocument: (This running inside the VM where API comes up okay) |
For me the latest containers are working. The CORS error is shown in the browser log even if the api container has an internal error. Do you have all containers running including dynamodb, S3 and the docker network? These are started if you are using |
|
Ran docker-compose up and then the s3-init script, thought that was all I was supposed to do; however I noticed last night the API server appears to be attempting to load its lambdas from the project directory on the Docker host. Those directories are empty, seems like I need to do some steps to build or need to fix the api server to reference pre-built lambdas within the image assuming there are any. To build the lambdas looks it would need aws-cdk and golang from the cdk folders json, ran out of time before I got a chance to scan the readmes |
|
Ah okay based on the |
|
Ah I got it. If you pull the latest docker images I published few days ago it should work, since I was building the images locally. The images before were broken because these have been built and published in the pipeline. I just introduced that pipeline. But I made a mistake. In the pipeline That's why it was empty. A pipeline step |
|
Nice finding! If you like you can raise a PR to fix it |
I'm not clear where |
|
here before the step |
|
Maybe you better add it, I see |
|
Oh ok sorry, I was blind, I added it already. Then I need to test this again after the next release, let's see |
|
Hello, The same issue is happening on my end. I installed it yesterday. A few seconds after the first ‘docker-compose’, the ‘rocketnotes-API’ exited with code 1, with these logs:
Does this PR fix something ? Maybe it hasn't accepted yet. Thank you for sharing your work, I hop I can try it soon. |
|
Nope, sadly my PR was just to fix the web container regression... API server still needs some kind of SAM fix, I don't know anything about SAM stuff, so just hoping for another release soonish so we can try again. |
|
Hi, thanks for reporting @s0lstice. I am still not able to reproduce this error on my machine. Any kind of system information would be helpful |
|
I'd offer to give you my current VM, but it's likely going to be ~40GB exported (I have a bunch of other unrelated OSS on it). I could try to reproduce a smaller Virtualbox VM next week that replicates the issue, if that would be of value. |
|
I have this issue on Debian 11 (VM) and Ubuntu 24.04.1 LTS, Is AWS CLI required on host ? The ‘dynamodb-init’ script uses it, and there are no warning messages if there is a failure. |
The API server doesn't start properly when doing
docker-compose up, full log at the bottom.I tried to poke around within the container after...
docker run --rm -it -p 3002:3002 --entrypoint /bin/bash fynnfluegge/rocketnotes-api:latestThere was no
dockercommand available in the path...The text was updated successfully, but these errors were encountered: