[ML] Do not make autoscaling decision when memory is undetermined#90259
Merged
dimitris-athanasiou merged 4 commits intoelastic:mainfrom Sep 26, 2022
Conversation
Collaborator
|
Pinging @elastic/ml-core (Team:ML) |
droberts195
approved these changes
Sep 22, 2022
droberts195
left a comment
There was a problem hiding this comment.
LGTM, although it would be nice to assert that we always set both node size and tier size or neither of them (if that's possible without making a test fail - I might be missing a valid situation where it can happen).
...gin/ml/src/main/java/org/elasticsearch/xpack/ml/autoscaling/MlMemoryAutoscalingCapacity.java
Outdated
Show resolved
Hide resolved
...ugin/ml/src/main/java/org/elasticsearch/xpack/ml/autoscaling/MlMemoryAutoscalingDecider.java
Outdated
Show resolved
Hide resolved
Contributor
Author
|
@elasticmachine update branch |
When a processor autoscaling decider was added in elastic#89645, an unwanted change of behaviour sneaked in. In particular, if we cannot determine required memory capacity, we previously returned `new AutoscalingDeciderResult(null)` where as we now return an autoscaling result with no memory capacity and whatever the result of the processor decider is. Previously, if we returned a result with null capacity, the cluster would remain as-is. Now, it is possible to cause unwanted scaling. This commit fixes this by checking if the memory decider result was undetermined and returns an empty result if so. Also, some logging warnings have been added to pop up scenarios that shouldn't happen like when the memory tracker is not called by the master node or it has no memory estimate for anomaly detection or analytics jobs.
e930a74 to
c31dbad
Compare
davidkyle
reviewed
Sep 23, 2022
...ugin/ml/src/main/java/org/elasticsearch/xpack/ml/autoscaling/MlMemoryAutoscalingDecider.java
Outdated
Show resolved
Hide resolved
Contributor
Author
|
run elasticsearch-ci/part-1 |
dimitris-athanasiou
added a commit
to dimitris-athanasiou/elasticsearch
that referenced
this pull request
Sep 26, 2022
…astic#90259) When a processor autoscaling decider was added in elastic#89645, an unwanted change of behaviour sneaked in. In particular, if we cannot determine required memory capacity, we previously returned `new AutoscalingDeciderResult(null)` where as we now return an autoscaling result with no memory capacity and whatever the result of the processor decider is. Previously, if we returned a result with null capacity, the cluster would remain as-is. Now, it is possible to cause unwanted scaling. This commit fixes this by checking if the memory decider result was undetermined and returns an empty result if so. Also, some logging warnings have been added to pop up scenarios that shouldn't happen like when the memory tracker is not called by the master node or it has no memory estimate for anomaly detection or analytics jobs.
Collaborator
💚 Backport successful
|
elasticsearchmachine
pushed a commit
that referenced
this pull request
Sep 26, 2022
…0259) (#90335) When a processor autoscaling decider was added in #89645, an unwanted change of behaviour sneaked in. In particular, if we cannot determine required memory capacity, we previously returned `new AutoscalingDeciderResult(null)` where as we now return an autoscaling result with no memory capacity and whatever the result of the processor decider is. Previously, if we returned a result with null capacity, the cluster would remain as-is. Now, it is possible to cause unwanted scaling. This commit fixes this by checking if the memory decider result was undetermined and returns an empty result if so. Also, some logging warnings have been added to pop up scenarios that shouldn't happen like when the memory tracker is not called by the master node or it has no memory estimate for anomaly detection or analytics jobs.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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.
When a processor autoscaling decider was added in #89645, an unwanted change of behaviour sneaked in. In particular, if we cannot determine required memory capacity, we previously returned
new AutoscalingDeciderResult(null)where as we now return an autoscaling result with no memory capacity and whatever the result of the processor decider is.Previously, if we returned a result with null capacity, the cluster would remain as-is. Now, it is possible to cause unwanted scaling.
This commit fixes this by checking if the memory decider result was undetermined and returns an empty result if so.
Also, some logging warnings have been added to pop up scenarios that shouldn't happen like when the memory tracker is not called by the master node or it has no memory estimate for anomaly detection or analytics jobs.