Dynamic send

[allevo]

This PR implements a new method to send a message:

• batch_send follows the same flow as send
• every message is queued on a dedicated Tokio task and sent asynchronously
• if the internal producer is closed, the task stops to send
• remove batch_publishing_delay params (breaking)
• put integration tests under tests to avoid parallel execution. Parallel execution causes problems running the tests locally due to the limitation of the OS file description number.
• fix typo on method (publising_id -> publishing_id) (breaking)
• add an hardcoded timeout (1sec) to send_with_confirm

I introduced also a bin to calculate the latency performance to proof the improvement on low send rate. The bin creates a producer and a consumer. Every minute, 50 messages with the current timestamp are sent to RabbitMQ and the consumer store the metric. Every minute, a log is printed with the statistics.

Before this PR:

$ cargo run --release --bin perf-producer
stats: Stats { average_latency: 59.54067, messages_received: 3000 }
stats: Stats { average_latency: 62.840668, messages_received: 3000 }
stats: Stats { average_latency: 58.391335, messages_received: 3000 }
stats: Stats { average_latency: 58.668335, messages_received: 3000 }

After this PR:

$ cargo run --release --bin perf-producer
stats: Stats { average_latency: 5.1963334, messages_received: 3000 }
stats: Stats { average_latency: 5.195667, messages_received: 3000 }
stats: Stats { average_latency: 5.288, messages_received: 3000 }
stats: Stats { average_latency: 5.148667, messages_received: 3000 }

[codecov]

Codecov Report

Attention: Patch coverage is 42.40000% with 144 lines in your changes missing coverage. Please review.

Project coverage is 89.00%. Comparing base (be36db2) to head (38b956a).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/bin/perf-producer.rs	0.00%	132 Missing ⚠️
src/producer.rs	89.58%	10 Missing ⚠️
src/client/mod.rs	87.50%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #276      +/-   ##
==========================================
- Coverage   90.17%   89.00%   -1.17%     
==========================================
  Files          80       81       +1     
  Lines        7390     7541     +151     
==========================================
+ Hits         6664     6712      +48     
- Misses        726      829     +103     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[DanielePalaia]

As this file is doing perf-tests any possibility to move it in the tests folder? (maybe creating a folder like performances)

[allevo]

We have different possibilities:

• put it under bench folder where, typically, you put performance tests, commonly developed with criterion crate. Anyway, this actually isn't a benchmark in the same sense of criterion which invokes the function multiple time and measure the duration.
• put it under tests folder and threat it as a integration test. Anyway, as a test, the output is not shown to the developer without the appropriate flag (--nocapture).
• put it under bin and threat is as a binary, as in this PR. You can run it using cargo run --release --bin perf-producer and print the output easily.

Anyway, I don't have a strong opinion on that and I can follow a different solution. let me know how you want to proceed.

[DanielePalaia]

I don't have a strong opinion too. As a preference I would choose the bench folder. It can be also useful in case in the future we want to add other performance tests and group them together

[DanielePalaia]

Looks good to me! @wolf4ood If you want to do a final validation too!

[Gsantomaggio]

This is not relevant, but is there a way to increase the code coverage?

[Gsantomaggio]

Thanks a lot @allevo @DanielePalaia