Datagen fixes #2599
Merged
Datagen fixes #2599
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
Fixes #2528. Signed-off-by: Gerd Zellweger <mail@gerdzellweger.com>
Start from 0 for array values, increment on every generated value. Fixes #2545. Signed-off-by: Gerd Zellweger <mail@gerdzellweger.com>
If datagen has to create very large records that take a long time to construct, creating a fixed amount determined by min(rate, 10_000) before inserting it as a batch is problematic as reported in #2558. The fix is to have an additional timeout during batch creation, if we take longer than one second to create a batch we just insert what we have so far and start a new batch. Signed-off-by: Gerd Zellweger <mail@gerdzellweger.com>
We reduce the default form 10*batch_size to batch_size. I did some microbenchmarks and it didn't lead to measurable contention for a reasonably sized record (TestStruct2). I also added a configuration option so the user can tune this, especially when having to consider lateness. Fixes #2558. Signed-off-by: Gerd Zellweger <mail@gerdzellweger.com>
Fixes #2585. Signed-off-by: Gerd Zellweger <mail@gerdzellweger.com>
ryzhyk
approved these changes
Sep 27, 2024
| @@ -410,12 +432,17 @@ impl InputGenerator { | |||
| // If we have a low rate, it's necessary to adjust the batch-size down otherwise no inserts might be visible | |||
| // for a long time (until a batch is full). | |||
| let batch_size: usize = min(plan.rate.unwrap_or(u32::MAX) as usize, 10_000); | |||
| let per_thread_chunk: usize = 10 * batch_size; | |||
| let per_thread_chunk: usize = plan.worker_chunk_size.unwrap_or(batch_size); | |||
There was a problem hiding this comment.
I realize now that we still won't be able to use multiple workers if we want to control lateness, since there's no upper bound on how much apart workers can be, but I still think this is better without 10x.
There was a problem hiding this comment.
You can set worker_chunk_size to a very low number num e.g., 1. This assumes that all worker threads take roughly equal amount of time to create a batch of course which assumes you don't have a config with e.g. zipf that leads to huge variability in record creation time if you're creating nested values. Maybe we can think of better strategies that coordinate the times that get produced among all workers.
Otherwise a user can accidentially specify a date that can't be represented. Fixes #2594.
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.
Fixes
incrementstrategy doesn't work correctly on arrays #2545