perf: Add template validation caching #13633
Closed
Conversation
This file contains 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
jakkubu
force-pushed
the
add-validation-template-caching
branch
from
fc3060e
to
a678d10
Compare
During template validation k8s API is called for each templateRef. For complex workflows with many refs it creates huge overhead. Let's cache such templates Throughout codebase each wrapper is used only in context of single call. It is not guaranteed for future, so TTL mechanism was added with default TTL=1M Additionally WithCacheTTL enables overwriting this parameter. Signed-off-by: Jakub Buczak <[email protected]>
jakkubu
force-pushed
the
add-validation-template-caching
branch
from
a678d10
to
9df7abf
Compare
Benchmarking creation of workflows with multiple-ref templateSetupBranches:
Using fresh Argo server started with Before each tests following procedure were followed:
Benchmarking tool: hey. It runs command in parallel by default 200 times using 50 workers. Those values can be modified using:
Typical call: hey \
-n 200 -c 50 \
-m POST \
-disable-keepalive \
-T "application/json" \
-d '{
"serverDryRun": false,
"workflow": {
"metadata": {
"generateName": "curl-echo-test-",
"namespace": "argo-test"
},
"spec": {
"workflowTemplateRef": {"name": "20-echos"},
"arguments": {}
}
}
}' \
https://localhost:2746/api/v1/workflows/argo-testResults
AppendixWorkflowsWorkflowsapiVersion: argoproj.io/v1alpha1
kind: WorkflowTemplate
metadata:
name: echo-1
spec:
entrypoint: print-message
templates:
- name: print-message
inputs:
parameters:
- name: message
value: "hello world!"
container:
image: docker/whalesay
command: [cowsay]
args: ["{{inputs.parameters.message}}"]apiVersion: argoproj.io/v1alpha1
kind: WorkflowTemplate
metadata:
name: echo-2
spec:
entrypoint: print-message
templates:
- name: print-message
inputs:
parameters:
- name: message
value: "hello world!"
container:
image: docker/whalesay
command: [cowsay]
args: ["{{inputs.parameters.message}}"]apiVersion: argoproj.io/v1alpha1
kind: WorkflowTemplate
metadata:
name: 20-echos
spec:
entrypoint: start
activeDeadlineSeconds: 1800
templates:
- name: start
dag:
tasks:
- name: echo-1
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-1
- name: echo-2
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-2
- name: echo-3
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-3
- name: echo-4
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-4
- name: echo-5
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-5
- name: echo-6
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-6
- name: echo-7
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-7
- name: echo-8
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-8
- name: echo-9
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-9
- name: echo-10
templateRef:
name: echo-1
template: print-message
arguments:
parameters:
- name: message
value: echo-10
- name: echo-11
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-11
- name: echo-12
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-12
- name: echo-13
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-13
- name: echo-14
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-14
- name: echo-15
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-15
- name: echo-16
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-16
- name: echo-17
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-17
- name: echo-18
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-18
- name: echo-19
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-19
- name: echo-20
templateRef:
name: echo-2
template: print-message
arguments:
parameters:
- name: message
value: echo-20`hey` output: Caching ON
|
agilgur5
reviewed
Sep 26, 2024
| wrapper.cacheMu.Lock() | ||
| defer wrapper.cacheMu.Unlock() | ||
|
|
||
| val, ok := wrapper.cache[name] |
There was a problem hiding this comment.
rather than a manual cache, this could probably accept a cache as an argument.
for instance, the Controller already has a cache of all WorkflowTemplates in its Informer.
the Server does not have an Informer yet, but will likely eventually have a SQLite cache similar to the one for the Workflow List
the CLI generally has none of the above though
There was a problem hiding this comment.
It makes sense - I did this as a quick resolution.
Do you prefer to use the same pattern as with workflow list or just add template informer to the server?
There was a problem hiding this comment.
#13672 is possibly fine for now, but the SQLite cache will eventually be needed for other features for filtering for WorkflowTemplates. It doesn't seem to be an often used part of the UI though, so maybe it's fine to leave without for now and modify when needed.
There was a problem hiding this comment.
FYI: #13672 is ready for review!
|
I'm continuing the work in PR #13672 |
agilgur5
added
the
solution/superseded
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.
Motivation
Improve performance of creating workflows with complex templateRef structure.
During template validation k8s API is called for each templateRef. For complex workflows with many refs it creates huge overhead. Let's cache such templates
Connected to issue #7418
Modifications
Add caching for
WorkflowTemplatesandClusterWorkflowTemplateswith default TTL=1minVerification
I added test cases checking if caching is working + performed a benchmarks (results in a comment in this PR)