Skip to content
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

Package review PR #1

Open
wants to merge 73 commits into
base: test
Choose a base branch
from
Open

Package review PR #1

wants to merge 73 commits into from

Conversation

KarthikSubbarao
Copy link
Member

@KarthikSubbarao KarthikSubbarao commented Apr 29, 2024

This is a PR which merges current changes in unstable into an empty branch in order to help with a review code of the entire codebase.

@KarthikSubbarao KarthikSubbarao force-pushed the unstable branch 3 times, most recently from 61aa1fb to 0e4d4d1 Compare April 30, 2024 18:14
@KarthikSubbarao KarthikSubbarao changed the title Package review Package review PR May 1, 2024
@KarthikSubbarao KarthikSubbarao self-assigned this May 1, 2024
src/commands/bloom.rs Outdated Show resolved Hide resolved
src/commands/bloom.rs Outdated Show resolved Hide resolved
src/commands/bloom.rs Outdated Show resolved Hide resolved
Comment on lines 276 to 296
let mut result = Vec::new();
match value {
Some(bf) => {
for item in input_args.iter().take(argc).skip(idx) {
result.push(RedisValue::Integer(bf.add_item(item.as_slice())));
}
Ok(RedisValue::Array(result))
}
None => {
if nocreate {
return Err(RedisError::Str("ERR not found"));
}
let mut bf = BloomFilterType::new_reserved(fp_rate, capacity, expansion);
for item in input_args.iter().take(argc).skip(idx) {
result.push(RedisValue::Integer(bf.add_item(item.as_slice())));
}
match filter_key.set_value(&BLOOM_FILTER_TYPE, bf) {
Ok(_) => Ok(RedisValue::Array(result)),
Err(_) => Err(RedisError::Str(ERROR)),
}
}
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could we check the bloom filter exists or not first and then insert the data in a single flow. Don't like the code duplication around insertion.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I can create a separate function for multi adds and call it from both flows. I wanted to handle both through the same flow, however it will be out of reference scope and result in a moved error

pub struct BloomFilterType {
pub expansion: u32,
pub fp_rate: f32,
pub filters: Vec<BloomFilter>,
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does the underlying support scalable bloom filters or do we need to support this mechanism?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This particular library does not have auto scaling

https://docs.rs/bloomfilter/1.0.13/bloomfilter/struct.Bloom.html

src/commands/bloom_util.rs Outdated Show resolved Hide resolved
Comment on lines 74 to 79
let new_capacity = filter.capacity * self.expansion;
let mut new_filter = BloomFilter::new(self.fp_rate, new_capacity);
// Add item.
new_filter.bloom.set(item);
new_filter.num_items += 1;
self.filters.push(new_filter);
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Shouldn't we fail the request if the expansion is much higher than we can support?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

expansion really means expansion_rate. I can rename this variable if that would help clarify this.

I don't think there is a limit on the number of sub filters an object can have. But (if we want) we can define a config for this to either silently fail expansion (by allowing a set) beyond a limit of X sub filters per object OR explicitly return an error (without a set).

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

One other aspect we should handle this checking / rejecting based on memory overhead of every operation that creates a new BloomFilter object (BF.ADD, BF.MADD, BD.RESERVE, BF.INSERT, RDB Load). Before any of these operations, we should probably check the memory usage and reject the operations if there is not sufficient space. We have a mechanism to compute the estimated additional memory overhead per creation.

src/wrapper/bloom_callback.rs Show resolved Hide resolved
src/commands/bloom_data_type.rs Outdated Show resolved Hide resolved
KarthikSubbarao pushed a commit that referenced this pull request Aug 23, 2024
Signed-off-by: Viktor Söderqvist <[email protected]>
Co-authored-by: Madelyn Olson <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
… objects. Update RDB load/save

Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
Signed-off-by: Karthik Subbarao <[email protected]>
zackcam and others added 2 commits November 5, 2024 12:01
…ters and memory_bytes. Additionally updated drop and created tests around this new info handler (#19)

Signed-off-by: zackcam <[email protected]>
* add bf.load and bf.dump to support bloom filter aofrewrite.

Signed-off-by: wuranxx <[email protected]>

add version in BloomFilterType.
change bf.load behavior: bf.load can't override exists keys.

Signed-off-by: wuranxx <[email protected]>

* remove bf.dump.
add bloom filter encode decode  comment.
add unit test for decode error case.

Signed-off-by: wuranxx <[email protected]>

* 1. Currently, the version field is added to the array only encode.
2. Add metrics when decode bloomFilterType.
3. Add metrics test when load bloom filter from aof.

Signed-off-by: wuranxx <[email protected]>

* 1. Add filter bytes check and unit test.
2. Fixed some comments and returned error messages.

Signed-off-by: wuranxx <[email protected]>

* fix `test_bf_decode_when_bytes_is_exceed_limit_should_failed` unit test.

Signed-off-by: wuranxx <[email protected]>

* Add expansion maximum value check to filter decode.

Signed-off-by: wuranxx <[email protected]>

---------

Signed-off-by: wuranxx <[email protected]>
@KarthikSubbarao KarthikSubbarao force-pushed the unstable branch 2 times, most recently from e74b825 to 188835b Compare November 21, 2024 21:50
* Support for DEBUG DIGEST module data type callback

Signed-off-by: Nihal Mehta <[email protected]>

* Update test cases

Signed-off-by: Nihal Mehta <[email protected]>

* Move digest to wrapper

Signed-off-by: Nihal Mehta <[email protected]>

* Update tests

Signed-off-by: Nihal Mehta <[email protected]>

* Add more scenarios for debug test

Signed-off-by: Nihal Mehta <[email protected]>

* Clean code and add scenario for debug test

Signed-off-by: Nihal Mehta <[email protected]>

---------

Signed-off-by: Nihal Mehta <[email protected]>
zackcam and others added 2 commits December 3, 2024 22:02
…filter dependency to version 3.0 (#23)

* Updating how we create BloomFilter from rdb loads. BloomFilter vec now has capacity of filter we are loading from

Signed-off-by: zackcam <[email protected]>

* Updating bloomfilter dependency to version 3, fixing breaking changes as well

Signed-off-by: zackcam <[email protected]>

* Updating the digest changes to follow updated version of bloom. As well as removing unnecesary fields saved in rdb

Signed-off-by: zackcam <[email protected]>

* Update log in src/bloom/data_type.rs

Signed-off-by: KarthikSubbarao <[email protected]>

* Update comment in src/bloom/utils.rs

Signed-off-by: KarthikSubbarao <[email protected]>

* Clippy error in src/bloom/data_type.rs

Signed-off-by: KarthikSubbarao <[email protected]>

---------

Signed-off-by: zackcam <[email protected]>
Signed-off-by: KarthikSubbarao <[email protected]>
Co-authored-by: KarthikSubbarao <[email protected]>
nnmehta and others added 6 commits December 6, 2024 13:29
…r structs (#24)

* Updating defragmentation to defrag both bloomfiltertype and bloomfiler structs

Signed-off-by: zackcam <[email protected]>

* Draft: Extra debugging and more factors being defragged

Signed-off-by: zackcam <[email protected]>

* Updating defrag and tests to use cursors and make test more robust by getting hits

Signed-off-by: zackcam <[email protected]>

* Fixing merge conflicts

Signed-off-by: zackcam <[email protected]>

* Fixing merge conflicts with random seed. Updating defrag test to use mexists and add_items_till_capacity. Refactored metric tracking as well to reduce code repetition

Signed-off-by: zackcam <[email protected]>

---------

Signed-off-by: zackcam <[email protected]>
…commands contain the correct acl categories (#29)

Signed-off-by: zackcam <[email protected]>
…e properties on replica nodes to match the object on the primary (#32)

Signed-off-by: Karthik Subbarao <[email protected]>
zackcam and others added 4 commits December 12, 2024 18:30
…s get incremented in defrag test (#33)

Signed-off-by: zackcam <[email protected]>
…fail (#34)

* Deflaking tests to remove chance of false positives causing tests to fail

Signed-off-by: zackcam <[email protected]>

* Update tests/test_bloom_command.py

Co-authored-by: KarthikSubbarao <[email protected]>
Signed-off-by: zackcam <[email protected]>

---------

Signed-off-by: zackcam <[email protected]>
Co-authored-by: KarthikSubbarao <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

7 participants