transform: create smp and scheduling groups #16114
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
Add a scheduling group parameter for all the work executed on the queue. Signed-off-by: Tyler Rockwood <rockwood@redpanda.com>
Signed-off-by: Tyler Rockwood <rockwood@redpanda.com>
rockwotj
force-pushed
the
transform-sg
branch
2 times, most recently
from
65304bc
to
7d17b82
Compare
Signed-off-by: Tyler Rockwood <rockwood@redpanda.com>
When RPCs come in for transform requests, ensure that they are running on the transform scheduling groups. Node local service requests are assumed to be already running on the transform scheduling group, which is mostly true except for deploys. Signed-off-by: Tyler Rockwood <rockwood@redpanda.com>
Signed-off-by: Tyler Rockwood <rockwood@redpanda.com>
The transform/rpc subsystem makes cross shard calls when accessing the correct partition for a transform, use a `smp_group` to manage this, like other subsystems. Signed-off-by: Tyler Rockwood <rockwood@redpanda.com>
Signed-off-by: Tyler Rockwood <rockwood@redpanda.com>
rockwotj
force-pushed
the
transform-sg
branch
from
7d17b82
to
99a3ece
Compare
|
ducktape was retried in https://buildkite.com/redpanda/redpanda/builds/43815#018d1472-5af1-48b9-90a0-a32b262d8698 |
oleiman
approved these changes
Jan 17, 2024
src/v/resource_mgmt/smp_groups.h
Outdated
Comment on lines
56
to
64
| return destroy_smp_service_group(*_kafka) | ||
| .then([this] { return destroy_smp_service_group(*_raft); }) | ||
| .then([this] { return destroy_smp_service_group(*_cluster); }) | ||
| .then([this] { return destroy_smp_service_group(*_proxy); }); | ||
| co_await destroy_smp_service_group(*_kafka); | ||
| co_await destroy_smp_service_group(*_raft); | ||
| co_await destroy_smp_service_group(*_cluster); | ||
| co_await destroy_smp_service_group(*_proxy); | ||
| co_await destroy_smp_service_group(*_transform); |
There was a problem hiding this comment.
question: Is this just for style points, or is there a specific reason to prefer coro style over continuation style here?
There was a problem hiding this comment.
Totally style points. coro4lyfe
|
I'm going to merge this so that I can get it into an install pack (via nightly) for our cloud for my testing. @dotnwat if you want to take a look I am happy to send another round of patches to address any feedback. This is mostly just plumbing around some scheduling groups and smp groups. |
|
/backport v23.3.x |
|
Failed to create a backport PR to v23.3.x branch. I tried: |
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.
Introduce SMP and CPU scheduling groups for transforms.
The main motivation for this is to allow for internal metrics on CPU usage for the transform subsystem. While doing this change, I also see that we generally also have smp groups too, so I created one for the internal data path, where there are a few cross shard calls to produce.
Backports Required
Release Notes
Improvements