Generate events when nodes are approved/rejected #3772

rossjones · 2024-04-10T14:08:22Z

When nodes are approved or rejected, or connected/disconnected, the system may want to respond to these events by creating new evaluations, or cancelling previous tasks.

This PR adds an event emitter which allows implementations of NodeEventHandler to receive events that are one of:

NodeEventApproved
NodeEventRejected
NodeEventDeleted
NodeEventConnected
NodeEventDisconnected

all events are delivered in a goroutine, and so it is the recipients responsibility to handle the event quickly, or to create a new goroutine and return immediately.

Currently this PR delivers events to an event listener in the manager package which currently logs that it has received and event. In future, we will want to modify this behaviour to create new evaluations, or stop current executions based on the node info that is delivered with the event, the job types currently deployed, and the type of event. To enable this, the NodeEventListener is provided with a reference to the evaluation broker and the jobstore.

coderabbitai · 2024-04-10T14:08:27Z

Important

Auto Review Skipped

Auto reviews are disabled on this repository.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository.

To trigger a single review, invoke the @coderabbitai review command.

Tips

Chat

There are 3 ways to chat with CodeRabbit:

Review comments: Directly reply to a review comment made by CodeRabbit. Example:
- I pushed a fix in commit <commit_id>.
- Generate unit testing code for this file.
- Open a follow-up GitHub issue for this discussion.
Files and specific lines of code (under the "Files changed" tab): Tag @coderabbitai in a new review comment at the desired location with your query. Examples:
- @coderabbitai generate unit testing code for this file.
- @coderabbitai modularize this function.
PR comments: Tag @coderabbitai in a new PR comment to ask questions about the PR branch. For the best results, please provide a very specific query, as very limited context is provided in this mode. Examples:
- @coderabbitai generate interesting stats about this repository and render them as a table.
- @coderabbitai show all the console.log statements in this repository.
- @coderabbitai read src/utils.ts and generate unit testing code.
- @coderabbitai read the files in the src/scheduler package and generate a class diagram using mermaid and a README in the markdown format.

Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments.

CodeRabbit Commands (invoked as PR comments)

@coderabbitai pause to pause the reviews on a PR.
@coderabbitai resume to resume the paused reviews.
@coderabbitai review to trigger a review. This is useful when automatic reviews are disabled for the repository.
@coderabbitai resolve resolve all the CodeRabbit review comments.
@coderabbitai help to get help.

Additionally, you can add @coderabbitai ignore anywhere in the PR description to prevent this PR from being reviewed.

CodeRabbit Configration File (`.coderabbit.yaml`)

You can programmatically configure CodeRabbit by adding a .coderabbit.yaml file to the root of your repository.
Please see the configuration documentation for more information.
If your editor has YAML language server enabled, you can add the path at the top of this file to enable auto-completion and validation: # yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json

Documentation and Community

Visit our Documentation for detailed information on how to use CodeRabbit.
Join our Discord Community to get help, request features, and share feedback.
Follow us on X/Twitter for updates and announcements.

frrist · 2024-04-10T19:13:29Z

pkg/node/manager/events.go

+		case <-completedChan:
+			return nil
+		case <-ctx.Done():
+			return nil


Probably worth returning ctx.Err() here

Done in ff7828f

frrist · 2024-04-10T19:17:34Z

pkg/node/manager/events.go

+	for {
+		select {
+		case <-completedChan:
+			return nil
+		case <-ctx.Done():
+			return nil
+		case <-e.clock.After(waitDuration):
+			return fmt.Errorf("timed out waiting for %s event callbacks to complete", event.String())
+		}
+	}


we can remove this for loop since select will block until one of its cases can run.

Done in ff7828f, not sure why it was there in the first place, probably force of habit.

frrist · 2024-04-10T19:40:49Z

pkg/node/manager/events.go

+	for _, hlr := range e.callbacks {
+		wg.Add(1)
+		go func(handler NodeEventHandler) {
+			handler.HandleNodeEvent(ctx, info, event)


I am thinking it's worth passing a sub-context with a timeout to these methods instead of the parent context. This way we ensure that all the handlers receive the cancellation signal and can stop their execution, and (try to) prevent goroutine leaks on timeout. What do you think?

Good idea, will give it a try. The problem might be that the timeout will not be using the mock clock, and so is only really useful for cleaning up unfinished callbacks rather than being useful to replace the timeout on each callback.

Cancellable version in 1da8eb2

The problem might be that the timeout will not be using the mock clock, and so is only really useful for cleaning up unfinished callbacks rather than being useful to replace the timeout on each callback

oo yeah, good point. Change looks good here - thanks.

frrist · 2024-04-10T19:41:45Z

pkg/node/manager/events.go

+		wg.Wait()
+	}()
+
+	waitDuration := 1 * time.Second


Can we make this a field on the NodeEventEmitter? It will be easier to change in the event we need to adjust this value later.

Done in ff7828f

frrist · 2024-04-10T19:44:04Z

pkg/node/manager/interfaces.go

+type NodeEventHandler interface {
+	HandleNodeEvent(ctx context.Context, info models.NodeInfo, event NodeEvent)
+}


I would find the code slightly more readable if we declared this interface where it's expected/used before the RegisterHandler method on the NodeEventEmitter instead of in a separate file.

Reasonble, done in ff7828f. I think I put it in an interfaces.go as there's been a bit of a pattern with the eval stuff that there's an interfaces.go and a types.go.

frrist · 2024-04-10T19:46:19Z

pkg/orchestrator/evaluation/inmemory_broker.go

@@ -815,3 +816,10 @@ func (b *InMemoryBroker) registerMetrics() (metric.Registration, error) {
 	}, orchestrator.EvalBrokerReady, orchestrator.EvalBrokerInflight, orchestrator.EvalBrokerPending,
 		orchestrator.EvalBrokerWaiting, orchestrator.EvalBrokerCancelable)
 }
+
+func (b *InMemoryBroker) HandleNodeEvent(ctx context.Context, info models.NodeInfo, evt manager.NodeEvent) {


For readability add a compile time assertion to the top of this file:
var _ manager.NodeEventHandler = &InMemoryBroker{}

Done in ff7828f

frrist · 2024-04-10T19:47:45Z

pkg/models/node_state.go

@@ -69,7 +69,6 @@ func (t liveness) IsValid() bool {
 type livenessContainer struct {
 	CONNECTED    NodeState
 	DISCONNECTED NodeState
-	HEALTHY      NodeState


CONNECTED now implies healthy I assume? or was this just never used?

It was never used and I think only worked because the default for ints is 0.

Correct "healthy" implied something that we don't know about the node. "Connected" is the right term (for now)

frrist · 2024-04-10T19:48:58Z

pkg/node/manager/events.go

+// WithClock is an option that can be used to set the clock for the NodeEventEmitter. This is useful
+// for testing purposes.
+func WithClock(clock clock.Clock) NodeEventEmitterOption {
+	return func(emitter *NodeEventEmitter) {
+		emitter.clock = clock
+	}
+}


Not sure this is used anywhere, was it meant to be used in a test?

It was, and is with b9a6953

frrist · 2024-04-11T15:51:39Z

pkg/node/manager/events_test.go

+	go func() {
+		s.clock.Add(10 * time.Second)
+	}()


why does this need a go routine? I'd assume adding 10 seconds to the clock is non-blocking?

Some weirdness with the mock clock where it won't work when it's inline. I've been trying to work out if I'm just holding it wrong but I suspect it's something to do with the work in EmitEvent happening in a goroutine and requiring a yield (but I'm not 100% sure)

frrist · 2024-04-11T23:15:53Z

@rossjones I have implemented a fix for the NodeInfo overwrite bug here: #3785
It was a very large mechanical change to make and sadly touched a lot of files. If it looks good to you, could we land your change (this PR) after it to avoid conflicts? I suspect it will be easier to resolved conflicts from your change here than the other way around. However if you spot any significant issues with #3785 go ahead and merge this and I'll sort the conflicts on my end.

We want to deliver events to a component that is able to use the job store and the events delivered to interact with an evaluation broker. This comment implements that event listener.

wdbaruni · 2024-04-16T13:45:23Z

@frrist are you owning this PR now? what do we need to do to unblock it and merge it?

frrist · 2024-04-16T15:34:00Z

@wdbaruni yeah I can take this over.

rossjones marked this pull request as ready for review April 10, 2024 16:21

rossjones requested a review from a team April 10, 2024 16:33

frrist reviewed Apr 10, 2024

View reviewed changes

rossjones requested a review from a team April 11, 2024 11:36

frrist reviewed Apr 11, 2024

View reviewed changes

frrist approved these changes Apr 11, 2024

View reviewed changes

frrist mentioned this pull request Apr 11, 2024

The requester node attempts to schedule work on disconnected nodes resulting in the job never running #3784

Closed

rossjones added 9 commits April 12, 2024 09:17

Add node event emitter to concurrently run callback events

0632042

Add deleted/connected/disconnected events

9eddaf5

Hook up events to eval broker

13a3611

Switch to handler and mocks for testing

4529b4b

Handle connection/disconnection events

85cbacc

Changes from PR review

4ad5289

Actually use the mock Clock so we do not wait in test

256ef60

Add a timeout to the context to cleanup any remaining goroutines

836ba77

Deliver events to manager-based component

3b5ffbc

We want to deliver events to a component that is able to use the job store and the events delivered to interact with an evaluation broker. This comment implements that event listener.

rossjones force-pushed the node-events branch from b2d909f to 3b5ffbc Compare April 12, 2024 08:17

rossjones assigned frrist Apr 12, 2024

rossjones added the blocked label Apr 12, 2024

wdbaruni removed the st/blocked label Apr 22, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Generate events when nodes are approved/rejected #3772

Generate events when nodes are approved/rejected #3772

rossjones commented Apr 10, 2024 •

edited

coderabbitai bot commented Apr 10, 2024

Auto Review Skipped

Chat

CodeRabbit Commands (invoked as PR comments)

CodeRabbit Configration File (`.coderabbit.yaml`)

Documentation and Community

frrist Apr 10, 2024

rossjones Apr 10, 2024

frrist Apr 10, 2024

rossjones Apr 10, 2024

frrist Apr 10, 2024

rossjones Apr 10, 2024 •

edited

frrist Apr 11, 2024

frrist Apr 10, 2024

rossjones Apr 10, 2024

frrist Apr 10, 2024

rossjones Apr 10, 2024

frrist Apr 10, 2024

rossjones Apr 10, 2024

frrist Apr 10, 2024

rossjones Apr 10, 2024

aronchick Apr 11, 2024

frrist Apr 10, 2024

rossjones Apr 10, 2024

frrist Apr 11, 2024

rossjones Apr 11, 2024

frrist commented Apr 11, 2024

wdbaruni commented Apr 16, 2024

frrist commented Apr 16, 2024

Generate events when nodes are approved/rejected #3772

Are you sure you want to change the base?

Generate events when nodes are approved/rejected #3772

Conversation

rossjones commented Apr 10, 2024 • edited

coderabbitai bot commented Apr 10, 2024

Auto Review Skipped

Chat

CodeRabbit Commands (invoked as PR comments)

CodeRabbit Configration File (.coderabbit.yaml)

Documentation and Community

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

rossjones Apr 10, 2024 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

frrist commented Apr 11, 2024

wdbaruni commented Apr 16, 2024

frrist commented Apr 16, 2024

rossjones commented Apr 10, 2024 •

edited

CodeRabbit Configration File (`.coderabbit.yaml`)

rossjones Apr 10, 2024 •

edited