filter_throttle: support retain messages if rate limit is exceeded

[wangzhuzhen]

This patch is to support retain messages if rate limit is exceeded.
https://github.com/fluent/fluent-bit/blob/master/plugins/filter_throttle/throttle.c#L84-L87
https://github.com/fluent/fluent-bit/blob/master/plugins/filter_throttle/throttle.c#L108-L115
https://github.com/fluent/fluent-bit/blob/master/plugins/filter_throttle/throttle.c#L212-L213
https://github.com/fluent/fluent-bit/blob/master/plugins/filter_throttle/throttle.c#L240-L241
https://github.com/fluent/fluent-bit/blob/master/plugins/filter_throttle/throttle.c#L247-L252
https://github.com/fluent/fluent-bit/blob/master/plugins/filter_throttle/throttle.c#L302-L306
https://github.com/fluent/fluent-bit/blob/master/plugins/filter_throttle/throttle.h#L35
https://github.com/fluent/fluent-bit/blob/master/plugins/filter_throttle/throttle.h#L42

For now, if rate limit is exceeded, throttle will drop the messages.
For now, if before fluent-bit first running, there is a input with huge messages which exceeded the throttle's (window * rate * interval), then only the first (window * rate * interval) records will be save, the others before fluent-bit running will be dropped.

Fixes #5356

Enter [N/A] in the box, if an item is not applicable to your change.

Testing
Before we can approve your change; please submit the following in a comment:

• Example configuration file for the change
• Debug log output from testing the change

• Attached Valgrind output that shows no leaks or memory corruption was found

Configuration

[SERVICE]
    Log_Level     info
[INPUT]
    name tail
    tag  xxx
    path /home/wong/fluent-bit-test/input.log
    DB   /home/wong/fluent-bit-test/flb.db
    read_from_head true
    Static_Batch_Size 50M

[FILTER]
    Name     throttle
    Match    *
    Rate     1000
    Window   15
    Interval 1s
    Print_Status true 
    Retain   false

[OUTPUT]
    name  file
    match *
    path  /home/wong/fluent-bit-test/output.log

For file /home/wong/fluent-bit-test/input.log with 150,000 lines records with format:

"[$id] aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"

we can check the result in output file /home/wong/fluent-bit-test/output.log by timestamp and account lines

Debug log

# fluent-bit  -c fluent-bit.conf
Fluent Bit v1.9.3
* Copyright (C) 2015-2022 The Fluent Bit Authors
* Fluent Bit is a CNCF sub-project under the umbrella of Fluentd
* https://fluentbit.io

[2022/04/25 16:40:33] [ info] [fluent bit] version=1.9.3, commit=, pid=23178
[2022/04/25 16:40:33] [ info] [storage] version=1.1.6, type=memory-only, sync=normal, checksum=disabled, max_chunks_up=128
[2022/04/25 16:40:33] [ info] [cmetrics] version=0.3.1
[2022/04/25 16:40:33] [ info] [filter:throttle:throttle.0] 1650876033: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 0 per interval
[2022/04/25 16:40:33] [ info] [sp] stream processor started
[2022/04/25 16:40:33] [ info] [output:file:file.0] worker #0 started
[2022/04/25 16:40:34] [ info] [filter:throttle:throttle.0] 1650876034: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 1000 per interval
[2022/04/25 16:40:35] [ info] [filter:throttle:throttle.0] 1650876035: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 1000 per interval
[2022/04/25 16:40:36] [ info] [filter:throttle:throttle.0] 1650876036: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 1000 per interval
[2022/04/25 16:40:37] [ info] [filter:throttle:throttle.0] 1650876037: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 1000 per interval
[2022/04/25 16:40:38] [ info] [filter:throttle:throttle.0] 1650876038: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 1000 per interval
[2022/04/25 16:40:39] [ info] [filter:throttle:throttle.0] 1650876039: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 1000 per interval
[2022/04/25 16:40:40] [ info] [filter:throttle:throttle.0] 1650876040: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 1000 per interval
[2022/04/25 16:40:41] [ info] [filter:throttle:throttle.0] 1650876041: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 1000 per interval
[2022/04/25 16:40:42] [ info] [filter:throttle:throttle.0] 1650876042: limit is 1000.00 per 1s with window size of 15, retain is 1, current rate is: 1000 per interval
....

If this is a change to packaging of containers or native binaries then please confirm it works for all targets.

• [ N/A] Attached local packaging test output showing all targets (including any new ones) build.

Documentation

• Documentation required for this feature

fluent/fluent-bit-docs#796

Backporting

• [ N/A] Backport to latest stable release.

---

Fluent Bit is licensed under Apache 2.0, by submitting this pull request I understand that this code will be released under the terms of that license.

[wangzhuzhen]

@koleini @fujimotos @leonardo-albertovich @edsiper Please review this PR and give some feedbacks, Thank you very much!

[github-actions]

This PR is stale because it has been open 45 days with no activity. Remove stale label or comment or this will be closed in 10 days.

[github-actions]

This PR is stale because it has been open 45 days with no activity. Remove stale label or comment or this will be closed in 10 days.

[github-actions]

This PR is stale because it has been open 45 days with no activity. Remove stale label or comment or this will be closed in 10 days.

[eschabell]

@patrick-stephens can you maybe review this PR?

[eschabell]

The docs-required tag can be removed @patrick-stephens as the docs PR fluent/fluent-bit-docs#796 is waiting.

[patrick-stephens]

Can you clarify the use case?

Why would you use a throttle plugin that retains data instead of dropping it? Just don't use it if you want to retain data?

[patrick-stephens]

This function seems redundant, all it does is return the value of ctx->retain_data