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

Require exporter timeouts to be positive #6850

Open
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

jack-berg
Copy link
Member

Resolves #6245.

This is a behavior change, which I would consider a bug. The behavior when exporter timeout 0 is at best ambiguous, and at worst non-functional. As discussed in this comment:

There's different interpretations of 0 for different http clients:

  • okhttp allows you to set the connect timeout to 0 with "no timeout" semantics.
  • jdk http client throws if you set connect timeout to 0

I figured we get around this by only accepting positive values. If a user wants forever semantics, they can se timeout to be Long.MAX_VALUE.

Configuration options impacted:

  • OtlpHttpSpanExporterBuilder#setTimeout
  • OtlpHttpSpanExporterBuilder#setConnectTimeout
  • OtlpHttpMetricExporterBuilder#setTimeout
  • OtlpHttpMetricExporterBuilder#setConnectTimeout
  • OtlpHttpLogRecordExporterBuilder#setTimeout
  • OtlpHttpLogRecordExporterBuilder#setConnectTimeout
  • OtlpGrpcSpanExporterBuilder#setTimeout
  • OtlpGrpcSpanExporterBuilder#setConnectTimeout
  • OtlpGrpcMetricExporterBuilder#setTimeout
  • OtlpGrpcMetricExporterBuilder#setConnectTimeout
  • OtlpGrpcLogRecordExporterBuilder#setTimeout
  • OtlpGrpcLogRecordExporterBuilder#setConnectTimeout
  • BatchLogRecordProcessorBuilder#setExporterTimeout
  • BatchSpanProcessorBuilder#setExporterTimeout

@jack-berg jack-berg requested a review from a team as a code owner November 5, 2024 19:49
Copy link

codecov bot commented Nov 5, 2024

Codecov Report

Attention: Patch coverage is 80.00000% with 3 lines in your changes missing coverage. Please review.

Project coverage is 90.28%. Comparing base (a6b3302) to head (5ad8be7).

Files with missing lines Patch % Lines
...er/otlp/logs/OtlpGrpcLogRecordExporterBuilder.java 50.00% 0 Missing and 1 partial ⚠️
...er/otlp/metrics/OtlpGrpcMetricExporterBuilder.java 50.00% 0 Missing and 1 partial ⚠️
...porter/otlp/trace/OtlpGrpcSpanExporterBuilder.java 50.00% 0 Missing and 1 partial ⚠️
Additional details and impacted files
@@            Coverage Diff            @@
##               main    #6850   +/-   ##
=========================================
  Coverage     90.28%   90.28%           
  Complexity     6588     6588           
=========================================
  Files           729      729           
  Lines         19768    19768           
  Branches       1944     1944           
=========================================
+ Hits          17847    17848    +1     
  Misses         1327     1327           
+ Partials        594      593    -1     

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

Copy link
Member

@trask trask left a comment

Choose a reason for hiding this comment

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

to make sure we're interpreting "timeout zero" consistently, would you support a similar spec clarification to

  • OTEL_EXPORTER_OTLP_TIMEOUT
  • OTEL_EXPORTER_ZIPKIN_TIMEOUT
  • OTEL_BSP_EXPORT_TIMEOUT
  • OTEL_BLRP_EXPORT_TIMEOUT
  • OTEL_METRIC_EXPORT_TIMEOUT

@jack-berg
Copy link
Member Author

to make sure we're interpreting "timeout zero" consistently, would you support a similar spec clarification to

Yes, that's a good call. But I wonder if there's a scenario where the outcome of a spec PR would influence what we do.

  • If there's not consistency on whether timeout of 0 is valid or invalid and spec rejects any attempt to clarify, then I think we're ok to interpret it as invalid in java.
  • If there's consistency that timeout of 0 is invalid and that clarifying is a bug fix, then we're ok with this PR in java.

I think the only situation where we wouldn't want to this PR in java would be: There is spec consensus that timeout of 0 conveys "indefinite". Seems hard to believe that all the implementations interpret it this way.

@jack-berg
Copy link
Member Author

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.

OTLP timeout validation should require positive values (> 0) instead of non-negative (>= 0)
2 participants