feat: implement HostDetector

[rauno56]

Which problem is this PR solving?

There's currently no detector for host.* resource attributes.

Short description of the changes

Added a new detector for detecting host attributes which are not cloud-based:

• host.name and
• host.arch.

Type of change

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

How Has This Been Tested?

• Added unit tests

Checklist:

• Followed the style guidelines of this project
• Unit tests have been added
• Documentation has been updated

[blumamir]

Thanks @rauno56 , The HostDetector.ts file is currently not committed in this PR.

Do you think it should be hosted in core repo or contrib detectors?

[rauno56]

Do you think it should be hosted in core repo or contrib detectors?

If it does, so do ProcessDetector imho. I see that the contrib detectors are vendor specific for now and a similar(ProcessDetector) one was here -> added this here as well.

[blumamir]

Do you think it should be hosted in core repo or contrib detectors?

If it does, so do ProcessDetector imho. I see that the contrib detectors are vendor specific for now and a similar(ProcessDetector) one was here -> added this here as well.

Sounds reasonable 👍

[codecov]

Codecov Report

Merging #2921 (981c0f1) into main (dc44b86) will decrease coverage by 0.03%.
The diff coverage is 31.57%.

❗ Current head 981c0f1 differs from pull request most recent head 91921be. Consider uploading reports for the commit 91921be to get more accurate results

@@            Coverage Diff             @@
##             main    #2921      +/-   ##
==========================================
- Coverage   91.85%   91.81%   -0.04%     
==========================================
  Files         176       70     -106     
  Lines        5573     2017    -3556     
  Branches     1209      441     -768     
==========================================
- Hits         5119     1852    -3267     
+ Misses        454      165     -289     

Impacted Files	Coverage Δ
...ntelemetry-resources/src/detectors/HostDetector.ts	23.52% <23.52%> (ø)
...ges/opentelemetry-resources/src/detectors/index.ts	100.00% <100.00%> (ø)
...s/opentelemetry-core/src/platform/node/sdk-info.ts	0.00% <0.00%> (-100.00%)	⬇️
...opentelemetry-core/src/platform/node/globalThis.ts	0.00% <0.00%> (-100.00%)	⬇️
...pentelemetry-core/src/platform/node/performance.ts	0.00% <0.00%> (-100.00%)	⬇️
...emetry-api-metrics/src/platform/node/globalThis.ts	0.00% <0.00%> (-100.00%)	⬇️
...lemetry-resources/src/detectors/ProcessDetector.ts	31.81% <0.00%> (-63.64%)	⬇️
...ckages/opentelemetry-exporter-zipkin/src/zipkin.ts	84.48% <0.00%> (-15.52%)	⬇️
...strumentation/src/platform/node/instrumentation.ts
...telemetry-sdk-trace-node/src/NodeTracerProvider.ts
... and 121 more

[rauno56]

Another question is whether it needs to be a separate detector in the first place.
Sure, it does map to spec better this way, and as long it's included in a package with ProcessDetector there's no (significant) harm done, but as an user I wouldn't want to install 3 different detectors for process, os, and host attributes for no good reason.

[blumamir]

Another question is whether it needs to be a separate detector in the first place. Sure, it does map to spec better this way, and as long it's included in a package with ProcessDetector there's no (significant) harm done, but as an user I wouldn't want to install 3 different detectors for process, os, and host attributes for no good reason.

I guess people might not always want all the attributes and separating them into different detectors can help use just the attributes needed.
But I can also see how those 3 detectors collect very similar data and might be more useful as one.
An alternative could be to define a composite detector that invokes those 3 detectors and merges the result into a single resource.

[rauno56]

An alternative could be to define a composite detector that invokes those 3 detectors and merges the result into a single resource.

Current SDK uses an utility for that.

I guess people might not always want all the attributes and separating them into different detectors can help use just the attributes needed.

I agree, but that might also happen among attributes detected from a single detector. One might only care about PID from the ProcessDetector but find process.command_line mostly dead weight, ergo we'd likely want to enable users to strip away attributes they don't need anyways. It's currently possible with wrapping the detectors with something like the utility mentioned ☝️

[blumamir]

I agree, but that might also happen among attributes detected from a single detector. One might only care about PID from the ProcessDetector but find process.command_line mostly dead weight, ergo we'd likely want to enable users to strip away attributes they don't need anyways. It's currently possible with wrapping the detectors with something like the utility mentioned ☝️

Yeah, good points. Might be worth thinking about how to offer such capabilities in the future.
Anyhow, it's not relevant to this PR.
LGTM. Thanks for working on it 💯

[rauno56]

Hmm... changing changelog manually creates a situation with a lot of conflict resolution.

I now finally understand the benefit of changefiles.