Profiling improvements #1767
Merged
Profiling improvements #1767
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
A trivial typo in the implementation of LeanVec resulted in completely incorrect reporting of heap usage in operator profiles. Signed-off-by: Leonid Ryzhyk <leonid@feldera.com>
|
@blp , one potential use of this is that we should be able to see where memory is still used in the nexmark benchmark running with storage. |
This commit switches to using jemalloc as our default allocator, enables its continuous profiling feature, and adds an HTTP endpoint to extract the heap profile at runtime. It's based on this blog: https://www.polarsignals.com/blog/posts/2023/12/20/rust-memory-profiling In more detail: The jemalloc allocator has a built-in profiler that can be used to continuously track heap usage of the program with low overhead, to the extent that it can be kept always on (at least that's the theory). It outputs the profile in a format compatible with the `pprof` tool from Google. I tried it with a release build of a pipeline and it generates perfectly legible profiles. I switched to using jemalloc as the global allocator with profiling enabled unconditionally for all pipelines. This way we will be able to get the profile of any pipeline. Signed-off-by: Leonid Ryzhyk <leonid@feldera.com>
added 2 commits
May 18, 2024 16:05
Signed-off-by: Leonid Ryzhyk <leonid@feldera.com>
The test applied integrate_trace_retain_keys to unspilled batches, so it didn't have the expected effect, but the test did not properly check this. Signed-off-by: Leonid Ryzhyk <leonid@feldera.com>
mihaibudiu
approved these changes
May 19, 2024
| @@ -1455,8 +1455,8 @@ impl<T: SizeOf> SizeOf for LeanVec<T> { | |||
| fn size_of_children(&self, context: &mut Context) { | |||
| self.vec.size_of_children(context, &mut |x, ctx| { | |||
| let x = unsafe { &*(x as *const T) }; | |||
| x.size_of_children(ctx); | |||
| }) | |||
| T::size_of_children(x, ctx); | |||
There was a problem hiding this comment.
I am not sure we have properly implemented this trait everywhere. We should make an audit
| //println!("retain_vals: {}bytes", trace.size_of().total_bytes()); | ||
| assert!(trace.size_of().total_bytes() < 50000); | ||
| // println!("retain_vals: {}bytes", trace.size_of().total_bytes()); | ||
| assert!(trace.size_of().total_bytes() < 70000); |
There was a problem hiding this comment.
Are these changes for to the bug you fixed?
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.
Is this a user-visible change (yes/no): yes