fix(llmobs): stop the Bedrock token-header cache growing without bound

[mz1119]

Last in the stack: #10436 → #10437 → this. Its base moves to master once #10437 lands.

What does this PR do?

Bedrock caches a token-header record per response attempt, keyed by request id, and releases it when :complete: fires. Two problems:

• Nothing bounded the cache, so entries no :complete: ever claims (a retried or aborted attempt) accumulate for the life of the process.
• A failed request never released its entry at all. The rejection path builds the response as { request, error, ...result } with no result, so there is no top-level $metadata and the request id has to be read off error.$metadata.

Bounded at 1000 entries with oldest-first eviction, and the id is now read from either place.

Motivation

Cache hygiene, unrelated to the gen_ai.* work and not urgent, which is why it sits last in the stack rather than under it.

🤖 Generated with Claude Code

[dd-octo-sts]

Overall package size

Self size: 9.22 MB
Deduped: 9.95 MB
No deduping: 9.95 MB

Dependency sizes

| name | version | self size | total size | |------|---------|-----------|------------| | import-in-the-middle | 3.5.1 | 127.66 kB | 513.61 kB | | opentracing | 0.14.7 | 194.81 kB | 194.81 kB | | dc-polyfill | 0.1.11 | 25.74 kB | 25.74 kB |

_{🤖 This report was automatically generated by heaviest-objects-in-the-universe}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.80%. Comparing base (7b2c66c) to head (7e50b23).

Additional details and impacted files

@@                              Coverage Diff                              @@
##           max.zhang/llmobs-gen-ai-apm-tags-disabled   #10452      +/-   ##
=============================================================================
- Coverage                                      98.81%   98.80%   -0.01%     
=============================================================================
  Files                                           1048     1048              
  Lines                                         167192   167003     -189     
  Branches                                          74       74              
=============================================================================
- Hits                                          165203   165014     -189     
  Misses                                          1989     1989              

Flag	Coverage Δ
ai-guard	64.11% <ø> (+0.01%)	⬆️
apm-capabilities	63.16% <58.62%> (+0.04%)	⬆️
apm-integrations	81.21% <ø> (+0.02%)	⬆️
appsec	58.02% <ø> (+<0.01%)	⬆️
debugger	69.94% <ø> (-0.01%)	⬇️
instrumentation	54.83% <ø> (+<0.01%)	⬆️
llmobs	80.39% <100.00%> (-0.01%)	⬇️
master-coverage	?
openfeature	65.91% <ø> (-0.18%)	⬇️
platform	68.78% <ø> (-0.03%)	⬇️
profiling	66.45% <ø> (+0.02%)	⬆️
serverless	65.58% <100.00%> (+0.03%)	⬆️
test-optimization	82.00% <ø> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[datadog-datadog-us1-prod]

🎉 All green!

🧪 All tests passed
❄️ No new flaky tests detected

🔄 Datadog retried 1 test - 1 passed on retry

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 98.26% (+0.02%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 7e50b23 | Docs | View more details | Give us feedback!}

[pr-commenter]

Benchmarks

Benchmark execution time: 2026-09-24 18:15:57

Comparing candidate commit 7e50b23 in PR branch max.zhang/llmobs-enabled-path-fixes with baseline commit 7b2c66c in branch max.zhang/llmobs-gen-ai-apm-tags-disabled.

📊 Benchmarking dashboard

Found 0 performance improvements and 0 performance regressions! Performance is the same for 2303 metrics, 10 unstable metrics.

Explanation

This is an A/B test comparing a candidate commit's performance against that of a baseline commit. Performance changes are noted in the tables below as:

• 🟩 = significantly better candidate vs. baseline
• 🟥 = significantly worse candidate vs. baseline

We compute a confidence interval (CI) over the relative difference of means between metrics from the candidate and baseline commits, considering the baseline as the reference.

If the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD), the change is considered significant.

Feel free to reach out to #apm-benchmarking-platform on Slack if you have any questions.

More details about the CI and significant changes

You can imagine this CI as a range of values that is likely to contain the true difference of means between the candidate and baseline commits.

CIs of the difference of means are often centered around 0%, because often changes are not that big:

---------------------------------(------|---^--------)-------------------------------->
                              -0.6%    0%  0.3%     +1.2%
                                 |          |        |
         lower bound of the CI --'          |        |
sample mean (center of the CI) -------------'        |
         upper bound of the CI ----------------------'

As described above, a change is considered significant if the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD).

For instance, for an execution time metric, this confidence interval indicates a significantly worse performance:

----------------------------------------|---------|---(---------^---------)---------->
                                       0%        1%  1.3%      2.2%      3.1%
                                                  |   |         |         |
       significant impact threshold --------------'   |         |         |
                      lower bound of CI --------------'         |         |
       sample mean (center of the CI) --------------------------'         |
                      upper bound of CI ----------------------------------'

Unstable benchmarks

These benchmarks have a confidence interval too wide to call a change; treat them as noise rather than signal.

scenario:appsec-appsec-enabled-24

• unstable max_rss_usage [-40.455MB; +7.234MB] or [-13.711%; +2.452%]

scenario:appsec-iast-no-vulnerability-iast-enabled-default-config-20

• unstable max_rss_usage [-11938.639KB; +11758.639KB] or [-5.207%; +5.129%]

scenario:debugger-line-probe-without-snapshot-24

• unstable cpu_user_time [-765.876ms; +1076.469ms] or [-8.445%; +11.870%]
• unstable execution_time [-785.907ms; +1093.365ms] or [-7.696%; +10.706%]
• unstable throughput [-113.183op/s; +76.490op/s] or [-10.118%; +6.838%]

scenario:debugger-line-probe-without-snapshot-26

• unstable cpu_user_time [-375.512ms; +526.902ms] or [-4.398%; +6.171%]

scenario:encoders-0.4-immediate-flush-20

• unstable max_rss_usage [-2.177MB; +6.930MB] or [-2.571%; +8.183%]

scenario:plugin-elasticsearch-bulk-index-26

• unstable execution_time [-89.471ms; +139.239ms] or [-4.232%; +6.586%]

scenario:plugin-mongodb-core-binary-hash-20

• unstable execution_time [-136.506ms; +214.743ms] or [-4.470%; +7.033%]

scenario:plugin-mongodb-core-binary-hash-24

• unstable execution_time [-109.294ms; +204.115ms] or [-3.728%; +6.962%]