Fix issue with peer stream node cleanup. #17235
Merged
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
This commit encompasses a few problems that are closely related due to their proximity in the code. 1. The peerstream utilizes node IDs in several locations to determine which nodes / services / checks should be cleaned up or created. While VM deployments with agents will likely always have a node ID, agentless uses synthetic nodes and does not populate the field. This means that for consul-k8s deployments, all services were likely bundled together into the same synthetic node in some code paths (but not all), resulting in strange behavior. The Node.Node field should be used instead as a unique identifier, as it should always be populated. 2. The peerstream cleanup process for unused nodes uses an incorrect query for node deregistration. This query is NOT namespace aware and results in the node (and corresponding services) being deregistered prematurely whenever it has zero default-namespace services and 1+ non-default-namespace services registered on it. This issue is tricky to find due to the incorrect logic mentioned in #1, combined with the fact that the affected services must be co-located on the same node as the currently deregistering service for this to be encountered. 3. The stream tracker did not understand differences between services in different namespaces and could therefore report incorrect numbers. It was updated to utilize the full service name to avoid conflicts and return proper results.
hashi-derek
added
backport/1.14
backport/1.15
DanStough
approved these changes
May 8, 2023
This was referenced May 8, 2023
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.
This commit encompasses a few problems that are closely related due to their proximity in the code.
The peerstream utilizes node IDs in several locations to determine which nodes / services / checks should be cleaned up or created. While VM deployments with agents will likely always have a node ID, agentless uses synthetic nodes and does not populate the field. This means that for consul-k8s deployments, all services were likely bundled together into the same synthetic node in some code paths (but not all), resulting in strange behavior. The Node.Node field should be used instead as a unique identifier, as it should always be populated.
The peerstream cleanup process for unused nodes uses an incorrect query for node deregistration. This query is NOT namespace aware and results in the node (and corresponding services) being deregistered prematurely whenever it has zero default-namespace services and 1+ non-default-namespace services registered on it. This issue is tricky to find due to the incorrect logic mentioned in 1, combined with the fact that the affected services must be co-located on the same node as the currently deregistering service for this to be encountered.
The stream tracker did not understand differences between services in different namespaces and could therefore report incorrect numbers. It was updated to utilize the full service name to avoid conflicts and return proper results.