@kubernetes step decorator #488
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
oavdeev
changed the title
oavdeev
force-pushed
the
k8s-executor
branch
4 times, most recently
from
0742ea7 to
7ff6be8
Compare
6 tasks
sappier
reviewed
Apr 29, 2021
Comment on lines
43
to
92
| from kubernetes import client | ||
|
|
||
| # Configureate Pod template container | ||
| container = client.V1Container( | ||
| name=job_name, | ||
| image=image, | ||
| command=command, | ||
| env=[client.V1EnvVar(name=k, value=v) for k, v in env.items()], | ||
| ) | ||
| # Create and configurate a spec section | ||
| template = client.V1PodTemplateSpec( | ||
| metadata=client.V1ObjectMeta(labels={"app": "pi"}), | ||
| spec=client.V1PodSpec(restart_policy="Never", containers=[container]), | ||
| ) | ||
| # Create the specification of deployment | ||
| spec = client.V1JobSpec( | ||
| template=template, | ||
| backoff_limit=1, # retries handled by Metaflow | ||
| ttl_seconds_after_finished=600, | ||
| ) | ||
|
|
||
| # Instantiate the job object | ||
| job = client.V1Job( | ||
| api_version="batch/v1", | ||
| kind="Job", | ||
| metadata=client.V1ObjectMeta(name=job_name, labels=labels), | ||
| spec=spec, | ||
| ) |
There was a problem hiding this comment.
k8s python client accepts specifications written as dictionaries, then it looks identical to the k8s json spec. The job could be created like this:
job = {
"apiVersion": "batch/v1",
"kind": "Job",
"metadata": {
"name": job_name,
"labels": labels
},
"spec": {
"template": {
"metadata": {
"labels": {
"app": "pi",
},
},
"spec": {
"containers": [
{
"name": job_name,
"image": image,
"command": command,
"env": [{"name": k, "value": v} for k, v in env.items()],
},
],
"restartPolicy": "Never"
}
},
"backoffLimit": 1, # retries handled by Metaflow
"ttlSecondsAfterFinished": 600
}
}
There was a problem hiding this comment.
Nice, didn't know that. One thing I like about client. API is that VSCode/mypy complains when I make a typo in a parameter name. For example I always forget that "template" {...} bit when writing YAML/JSON.
oavdeev
force-pushed
the
k8s-executor
branch
2 times, most recently
from
502c6c4 to
9e96c68
Compare
|
closing in favor of #644 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Note that it is PR is on top of #504, since it relies on some mflog refactorings introduced there.
What does this do
Introduce a new decorator,
@kubernetesthat allows you to run specific steps on kubernetes as Jobs.What works
Basics work. You can decorate steps with
@kubernetesand they'll run on k8s cluster.--with kubernetesworks too. Logging works through mflog.You can specify a docker image as
@kubernetes(image="python:3.7"). If the image is not specified, it usesMETAFLOW_KUBERNETES_IMAGE_URI. If that is missing too, it'll use default python image. This behavior is essentially the same as@batchtoday.This also introduces a dependency on Kubernetes Python client (only needed if you use
@kubernetes).Configuration and auth
K8S creds
It will use the k8s creds that are configured in your environment. That is, this doesn't do anything special for k8s auth; if
kubectlworks in your environment, this will too.AWS creds
The code still uses S3, and assumes that AWS auth is somehow taken care of via your cluster configuration. That is, AWS creds are available either via the instance profile, or via IRSA.
What's left
@timeout@resources@environment