ref(store): Random Kafka partitioning for sessions #1194
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
jan-auer
commented
Feb 23, 2022
| @@ -673,8 +673,8 @@ impl KafkaMessage { | |||
| Self::Attachment(message) => message.event_id.0, | |||
| Self::AttachmentChunk(message) => message.event_id.0, | |||
| Self::UserReport(message) => message.event_id.0, | |||
| Self::Session(message) => message.session_id, | |||
| Self::Metric(_message) => Uuid::nil(), // TODO(ja): Determine a partitioning key | |||
| Self::Session(_message) => Uuid::nil(), // Explicit random partitioning for sessions | |||
There was a problem hiding this comment.
Note that this NIL is overwritten immediately after the match block.
There was a problem hiding this comment.
nit, can we note down the observed scenario that motivated this change?
There was a problem hiding this comment.
I was hoping in code it's enough to say that we explicitly want uniform distribution. The full reasoning is in the PR description, and given that it's a hypothesis I wasn't gonna put this in the code.
jjbayer
approved these changes
Feb 23, 2022
|
Asking for @fpacifici's blessing before merging this :) |
untitaker
approved these changes
Feb 23, 2022
| @@ -673,8 +673,8 @@ impl KafkaMessage { | |||
| Self::Attachment(message) => message.event_id.0, | |||
| Self::AttachmentChunk(message) => message.event_id.0, | |||
| Self::UserReport(message) => message.event_id.0, | |||
| Self::Session(message) => message.session_id, | |||
| Self::Metric(_message) => Uuid::nil(), // TODO(ja): Determine a partitioning key | |||
| Self::Session(_message) => Uuid::nil(), // Explicit random partitioning for sessions | |||
There was a problem hiding this comment.
nit, can we note down the observed scenario that motivated this change?
* master: ref(make): Simplify M1 exports in Makefile (#1206) fix(metrics): Wait for project states during aggregator shutdown (#1205) fix(test): Find librdkafka on Apple M1 (#1204) build: Bump sentry-relay in dev dependencies to 0.8.9 (#1202) ref(metrics): Tag backdated bucket creations in statsd (#1200) feat(metrics): Extract user satisfaction as tag (#1197) fix(statsd): Add new metric_type tag to existing metrics (#1199) fix: Apply clippy 1.59 suggestions (#1198)
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.
Instead of using the session ID as partition key for producing sessions, assign
a random partitioning key. This ensures more even distribution of sessions on
the Kafka topic, even if clients are sending many updates for the same session.
We have noticed intermittent imbalance on the partitions. The theory is that
clients are sending repeat updates for the same session (ID) when a high number
of errors are occurring within that session. Every session update would go to
the same partition unless we use random partitioning.
Snuba does not have logical constraints on the partitioning.