Replies: 5 comments 1 reply
-
Update - adding the following to
...and I have Now I'm digging for a way to "map" the container from |
Beta Was this translation helpful? Give feedback.
-
That is wierd: the image exists on this repo (see https://github.com/serverlessworkflow/synapse/pkgs/container/synapse%2Frunner) and we are able to successfully use it on different machines 👀 Can you try to pull the image manually to see if that solves your problem?
You are absolutly right, I forgot to mention that one should build the runner image before attempting to use the
You can do so by navigating to the dashboard (by default localhost:8080), authenticating using one of the tokens configured in the tokens file and going to the operators page, which is found under "services" menu item. Then, proceed to edit your operator, and set the image you want to use in You should by then be good to go: create a new workflow and start an instance using the UI or the CLI, and watch your operator startup a new instance. On a side note, I know the doc is a bit rough for the time being (any help/suggestion welcome) but the wiki might give you some good pointers, especially (and not so obviously) the architecture section, which documents all environment variables that can be used to set you up 😉 I hope this helps! PS: the container platform is not be be mistaken for the runtime. As a matter of fact, the former is the platform used by runners to spawn the containers configured by By default, runners are spawned by the Docker runtime (alternatives are Native, which should only be used for test purposes, and Kubernetes), and by default runners use the Docker container platform (the only alternative is Kubernetes). Note that you while it is easier to configure matching runtime and container platform, you are free to use any combination of those. |
Beta Was this translation helpful? Give feedback.
-
Absolutely, I starred both repos! I was able to pull the image manually, as you suggested... and now the operator does launch it, however being an x86 image it core-dumps rather than properly completing on my ARM machine: I guess for me building the runner image locally is the way to go (all the other locally built images seem to run pretty well). I will dig thru the rest of the docs/ENV vars you pointed out (thanks for the link) to see if I can figure out how to get it to choose the locally built image. BTW, the container does appear to do some things before dying: 2024-09-13 16:30:19 [23:30:19] info: Microsoft.Hosting.Lifetime[0] |
Beta Was this translation helpful? Give feedback.
-
Thank you for your help Charles - with a bit more wrangling here tonight I was able to get the local image compiled and running locally on MacOS ARM. New question - has there been any consideration or discussion of runner processes available on standby (without spin-up time) for hot-path kinds of processing? |
Beta Was this translation helpful? Give feedback.
-
Happy to hear you got it working! I hope the software is to your liking!
Not until now, no, but that's a great idea!!! Would you be so kind to create a new feature request issue? |
Beta Was this translation helpful? Give feedback.
-
Great project! This week I was very excited to discover ServerlessWorkflow and am planning to utilize the spec for the "Rules / Workers" side of my ambitious developer-friendly low-code framework.
Currently I'm also attempting to evaluate synapse as a near-term runtime for the compute/orchestration. After many hours of trial + error, digging thru the repo and configuring (did not find the other discussion on alpha config until I had figured out most of it myself 🤦♂️), I finally have most things running using the
docker_compose.build.yml
setup (The latest thing I had to sort out was copying + customizing theSYNAPSE_RUNTIME_DOCKER_SECRETS_DIRECTORY
env var into the .build namespaced file out of thedocker_compose.yml
)but now it's saying: No such image: ghcr.io/serverlessworkflow/synapse/runner:1.0.0-alpha2
I can't see any
runner
image config's in deployments/docker-compose tree, although I do note this DockerfileAssuming that the
runner
should be expected to work via docker compose, would someone with more knowledge of the system internals be able to weigh in with some guidance for thedocker_compose.build.yml
lines for the runner?Beta Was this translation helpful? Give feedback.
All reactions