feat(codecs): add support for protobuf decoding

[Daniel599]

Hi
addresses: #9744
please note some of the 'TODOs' where I would your feedback.

This PR adds support for choosing a protobuf decoder for a source(e.g kafka)
The user provides a protobuf desc file and a 'message type', each message within the topic should be of 'message type'

[bits-bot]

All committers have signed the CLA.

[netlify]

✅ Deploy Preview for vector-project ready!

Name	Link
🔨 Latest commit	e53532d
🔍 Latest deploy log	https://app.netlify.com/sites/vector-project/deploys/64c3faf27869c40008973103
😎 Deploy Preview	https://deploy-preview-18019--vector-project.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site configuration.

[netlify]

✅ Deploy Preview for vrl-playground ready!

Name	Link
🔨 Latest commit	e53532d
🔍 Latest deploy log	https://app.netlify.com/sites/vrl-playground/deploys/64c3faf2e3799c0007b16707
😎 Deploy Preview	https://deploy-preview-18019--vrl-playground.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site configuration.

[pront]

Thank you for submitting this work @Daniel599!

[Daniel599]

@pront can you please advise me what to do here?
I don't see any Deserializer which returns a Result in its new method

[pront]

So here, I would make new look like this:

pub fn new(message_descriptor: MessageDescriptor) -> Self {
   Self { message_descriptor }
}

This way new will always succeed.

Then do everything that can fail in a util:

fn get_message_descriptor(desc_file: String, message_type: String) -> Result<MessageDescriptor> 
{ ... }

This will also gives us more flexibility in how we create a MessageDescriptor.

[Daniel599]

I made the requested change, but I still had to do unwrap() before the new. is that what you meant? I think I`m missing something

[bruceg]

We should avoid the unwrap since it can turn a configuration problem into a crash. Instead we would need to pass the error up to the caller. In the case of the From implementation, that would need to be turned into a TryFrom, which can return a Result. At some point in the chain of callers there will be a method that already returns a Result and the error can be returned there, ideally with a context indicating why the error occurred.

[pront]

So here, I would make new look like this:

pub fn new(message_descriptor: MessageDescriptor) -> Self {
   Self { message_descriptor }
}

This way new will always succeed.

Then do everything that can fail in a util:

fn get_message_descriptor(desc_file: String, message_type: String) -> Result<MessageDescriptor> 
{ ... }

This will also gives us more flexibility in how we create a MessageDescriptor.

[pront]

Thank you again @Daniel599, this is starting to look great.

Only a few things left before we can get this in:

• I replied to the following questions:
  • what Kind to use
  • refactor new()
• Commit my suggestions. We can evaluate performance again after we ensure correctness. It can be another PR later, you already did a lot work in this one.
• Add test coverage for prost_reflect::Value::List and prost_reflect::Value::Map. Just to make sure we hit all those branches. I noticed in the docs that some functions like this one may panic.

[pront]

A try_from might be better here.

[pront]

@Daniel599 I tried to push a commit to fix the last remaining issues but I don't have permission to push. Can you merge this final-israel#1 to your branch?

[Daniel599]

@pront Do I need to run make generate-component-docs or should I rebase my branch?

[bruceg]

It's simpler to just use from here, which is what the above ends up calling anyways:

Suggested change

	Into::<ProtobufDeserializer>::into(self)
	ProtobufDeserializer::from(self)

[bruceg]

nit: I don't see that this method is actually being called anywhere, and it just calls the default method anyways, so it seems fairly superfluous.

[bruceg]

We should avoid the unwrap since it can turn a configuration problem into a crash. Instead we would need to pass the error up to the caller. In the case of the From implementation, that would need to be turned into a TryFrom, which can return a Result. At some point in the chain of callers there will be a method that already returns a Result and the error can be returned there, ideally with a context indicating why the error occurred.

[bruceg]

I think this can be done without a clone:

Suggested change

	let b = fs::read(desc_file.clone())
	let b = fs::read(Path::new(&desc_file))

[bruceg]

File names should be stored in PathBuf type instead of strings to avoid corner cases of illegal filenames.

[pront]

This is a great point. Changed the type.

[bruceg]

This function doesn't need to take ownership of either string, so these can both be just &str. Using an &Path for the filename would be even better (see comment above).

[bruceg]

This interface is changing (today even). See #18097

[pront]

Hi @Daniel599, please allow me to push commits to this PR (or just keep approving my commits like yesterday), I will help with the remaining details. I will address all @bruceg comments and try to get this merged today!

[Daniel599]

Hi @Daniel599, please allow me to push commits to this PR (or just keep approving my commits like yesterday), I will help with the remaining details. I will address all @bruceg comments and try to get this merged today!

Sorry but seems I cannot "allow edits" since it's an organization repository, see.
I`ll approve your PR.

[pront]

Will follow-up with another PR to refactor build() to return a Result and avoid unwrap.

[pront]

This is a great point. Changed the type.

[pront]

regenerated the docs

[github-actions]

Regression Detector Results

Run ID: 34f72bfe-3105-4e90-85fe-6a32487ef76a
Baseline: 8a2f8f6
Comparison: a06c711
Total vector CPUs: 7

Explanation

A regression test is an integrated performance test for vector in a repeatable rig, with varying configuration for vector. What follows is a statistical summary of a brief vector run for each configuration across SHAs given above. The goal of these tests are to determine quickly if vector performance is changed and to what degree by a pull request.

Because a target's optimization goal performance in each experiment will vary somewhat each time it is run, we can only estimate mean differences in optimization goal relative to the baseline target. We express these differences as a percentage change relative to the baseline target, denoted "Δ mean %". These estimates are made to a precision that balances accuracy and cost control. We represent this precision as a 90.00% confidence interval denoted "Δ mean % CI": there is a 90.00% chance that the true value of "Δ mean %" is in that interval.

We decide whether a change in performance is a "regression" -- a change worth investigating further -- if both of the following two criteria are true:

1. The estimated |Δ mean %| ≥ 5.00%. This criterion intends to answer the question "Does the estimated change in mean optimization goal performance have a meaningful impact on your customers?". We assume that when |Δ mean %| < 5.00%, the impact on your customers is not meaningful. We also assume that a performance change in optimization goal is worth investigating whether it is an increase or decrease, so long as the magnitude of the change is sufficiently large.

2. Zero is not in the 90.00% confidence interval "Δ mean % CI" about "Δ mean %". This statement is equivalent to saying that there is at least a 90.00% chance that the mean difference in optimization goal is not zero. This criterion intends to answer the question, "Is there a statistically significant difference in mean optimization goal performance?". It also means there is no more than a 10.00% chance this criterion reports a statistically significant difference when the true difference in mean optimization goal is zero -- a "false positive". We assume you are willing to accept a 10.00% chance of inaccurately detecting a change in performance when no true difference exists.

The table below, if present, lists those experiments that have experienced a statistically significant change in mean optimization goal performance between baseline and comparison SHAs with 90.00% confidence OR have been detected as newly erratic. Negative values of "Δ mean %" mean that baseline is faster, whereas positive values of "Δ mean %" mean that comparison is faster. Results that do not exhibit more than a ±5.00% change in their mean optimization goal are discarded. An experiment is erratic if its coefficient of variation is greater than 0.1. The abbreviated table will be omitted if no interesting change is observed.

Changes in experiment optimization goals with confidence ≥ 90.00% and |Δ mean %| ≥ 5.00%:

experiment	goal	Δ mean %	confidence
http_text_to_http_json	ingress throughput	+5.29	100.00%

Fine details of change detection per experiment.

experiment	goal	Δ mean %	Δ mean % CI	confidence
http_text_to_http_json	ingress throughput	+5.29	[+5.22, +5.35]	100.00%
otlp_http_to_blackhole	ingress throughput	+2.76	[+2.60, +2.92]	100.00%
splunk_hec_route_s3	ingress throughput	+2.38	[+2.24, +2.52]	100.00%
socket_to_socket_blackhole	ingress throughput	+2.36	[+2.31, +2.41]	100.00%
datadog_agent_remap_datadog_logs_acks	ingress throughput	+2.31	[+2.20, +2.42]	100.00%
syslog_log2metric_splunk_hec_metrics	ingress throughput	+2.15	[+2.06, +2.25]	100.00%
datadog_agent_remap_blackhole	ingress throughput	+1.78	[+1.69, +1.87]	100.00%
syslog_loki	ingress throughput	+1.64	[+1.55, +1.73]	100.00%
otlp_grpc_to_blackhole	ingress throughput	+1.47	[+1.37, +1.58]	100.00%
http_to_http_acks	ingress throughput	+1.27	[+0.03, +2.52]	80.92%
datadog_agent_remap_datadog_logs	ingress throughput	+0.99	[+0.87, +1.11]	100.00%
file_to_blackhole	egress throughput	+0.84	[-2.92, +4.61]	22.61%
http_to_s3	ingress throughput	+0.52	[+0.32, +0.72]	99.91%
enterprise_http_to_http	ingress throughput	+0.06	[+0.02, +0.09]	94.39%
http_to_http_noack	ingress throughput	+0.04	[-0.02, +0.10]	65.44%
http_to_http_json	ingress throughput	+0.00	[-0.04, +0.04]	4.17%
fluent_elasticsearch	ingress throughput	-0.00	[-0.00, +0.00]	8.84%
splunk_hec_to_splunk_hec_logs_acks	ingress throughput	-0.00	[-0.06, +0.06]	1.81%
splunk_hec_indexer_ack_blackhole	ingress throughput	-0.01	[-0.05, +0.03]	20.22%
splunk_hec_to_splunk_hec_logs_noack	ingress throughput	-0.02	[-0.06, +0.03]	34.94%
datadog_agent_remap_blackhole_acks	ingress throughput	-0.17	[-0.27, -0.06]	95.85%
syslog_humio_logs	ingress throughput	-0.81	[-0.88, -0.75]	100.00%
syslog_splunk_hec_logs	ingress throughput	-1.52	[-1.59, -1.44]	100.00%
syslog_regex_logs2metric_ddmetrics	ingress throughput	-1.79	[-2.08, -1.51]	100.00%
syslog_log2metric_humio_metrics	ingress throughput	-3.78	[-3.86, -3.70]	100.00%

[jszwedko]

Sorry, I didn't review this before but this desc_file option should be nested under a protobuf key like the gelf and json options you see here, are.

[jszwedko]

Opened #18111 to move it.

[jszwedko]

@neuronull is this OK?

[neuronull]

Thanks for the call out. I think this is OK for now. We don't have a serializer for protobuf yet, so there isn't any alternative really. This is similar to the OctetCounting framer in the function below.
When we get to the validation of components that operate over protobuf, we will have to add a task to implement a protobuf external resource. It's possible we'd have to implement a protobuf serializer at that time, to get full coverage of said components.

[jszwedko]

I know we said we'll follow up on this TODO but I think we should have resolved it before this was merged since otherwise we have to keep track of it to resolve before the next release. To that end I created this issue to track it: #18110 and put it in a new 0.32.0 milestone to make sure we resolve it prior to the next release.