-
Notifications
You must be signed in to change notification settings - Fork 551
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Modify perf CDT tests to use the im4gn.xlarge
instance type
#18556
Conversation
"acks": "all", | ||
"linger.ms": 1, | ||
"max.in.flight.requests.per.connection": 5, | ||
"batch.size": 131072, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Lets set this to 16KB? It's a lot more realisitic and avoids us running in larger forced batch sizes.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice catch, switching it to the default of 16384
.
"consumer_per_subscription": 200, | ||
"producers_per_topic": 200, | ||
"producer_rate": 150_000, | ||
"consumer_per_subscription": 1, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I guess a lot of consumers aren't really relevant for this test?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, the goal here is to stress the produce path as much as possible. So I set this to reduce to load from the consumers to the lowest possible.
"acks": "all", | ||
"linger.ms": 1, | ||
"max.in.flight.requests.per.connection": 10, | ||
"batch.size": 1024, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Just set to 1?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I keep forgetting we can do that. Switching it now.
"enable.idempotence": "false", | ||
"acks": "all", | ||
"linger.ms": 1, | ||
"max.in.flight.requests.per.connection": 10, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think we should really test we idempotence on as it's the more common codepath.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Fair enough, switching it over.
"producer_rate": 150_000, | ||
"consumer_per_subscription": 1, | ||
"producers_per_topic": 10, | ||
"producer_rate": 30_000, | ||
"message_size": 1024, | ||
"payload_file": "payload/payload-1Kb.data", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We could even use the 200 byte message size here or something
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let me see if I can push the producer rate a bit higher with smaller messages.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Unsurprisingly using smaller messages doesn't allow for a higher producer rate.
This PR modifies the existing perf CDT tests to use a much small 3x
im4gn.xlarge
. The hope of doing this is to allow us to run more regression tests at the same cost as the existing tests.Backports Required
Release Notes