Collect metrics optionally #245
Open
Conversation
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
gabotechs
changed the title
gabotechs
force-pushed
the
gabrielmusat/collect-metrics-optionally
branch
from
f653b10 to
5fc4b41
Compare
jayshrivastava
approved these changes
Dec 5, 2025
| pub shuffle_batch_size: usize, default = 8192 | ||
| /// Propagate collected metrics from all nodes in the plan across network boundaries | ||
| /// so that they can be reconstructed on the head node of the plan. | ||
| pub collect_metrics: bool, default = false |
Collaborator
There was a problem hiding this comment.
nit: defaulting to true might be better since that's how vanilla df works.
| let target_task = partition / partitions_per_task; | ||
| let target_partition = partition % partitions_per_task; | ||
|
|
||
| // TODO: this propagation should be automatic <link to issue> |
Collaborator
There was a problem hiding this comment.
I think this is missing a link. Do you want to create an issue in this repo as well?
| let stream = if retrieve_metrics { | ||
| MetricsCollectingStream::new(stream, metrics_collection_capture).left_stream() | ||
| } else { | ||
| stream.right_stream() |
Collaborator
There was a problem hiding this comment.
Do you mind explaining this / leaving a comment? I pulled your branch and I honestly cannot tell where the Either type comes from.
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Outdated suggestions cannot be applied.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.
While investigating #228, I tried disabling metrics collection to see if that was somehow contributing to the deadlock.
The conclusion is that it not only does not affect the deadlock, but also does not affect performance in any measurable way.
As I had this code anyways that enables/disables metrics collection optionally, and I now that you @jayshrivastava is something you were planning to do, I though I might as well just create a separate PR for this.
The
collect_metricsfield inDistributedConfignow needs to be accessible not only on the head node, but also in any other node involved in the query, which means that we now need to sendDistributedConfigover the wire as an option extension. While implementing that, I stepped with #247, which prevents us from sending a mutated config extension over the wire.I made a workaround in this PR for addressing the issue manually with a
manually_propagate_distributed_config(), but once the issue gets fixed, manual propagation is no longer needed.