Skip to content

docs(dotAI): document Bedrock model ID forms, credentials, and error mapping#36025

Open
sfreudenthaler wants to merge 1 commit into
dot-ai-bedrock-timeout-retriesfrom
dot-ai-bedrock-docs-validation
Open

docs(dotAI): document Bedrock model ID forms, credentials, and error mapping#36025
sfreudenthaler wants to merge 1 commit into
dot-ai-bedrock-timeout-retriesfrom
dot-ai-bedrock-docs-validation

Conversation

@sfreudenthaler
Copy link
Copy Markdown
Member

@sfreudenthaler sfreudenthaler commented Jun 5, 2026 •
edited
Loading

Stacked PR 3/3 into #35242 — base is dot-ai-bedrock-timeout-retries (#36024). Offered as optional help, @ihoffmann-dot.

Proposed Changes

  • Class-level javadoc on BedrockModelProviderStrategy + extended ProviderConfig Bedrock block. Every claim verified by live invocations against real Bedrock models (R&D account, today):
    • Model ID forms: inference-profile-prefixed (us.deepseek.r1-v1:0 — bare ID rejected) vs bare on-demand (openai.gpt-oss-120b-1:0, amazon.titan-embed-text-v2:0 — no profile exists; prefix would fail)
    • Error decoder: on-demand throughput isn't supported → use profile prefix · This model doesn't support the stopSequences field → langchain4j-bedrock < 1.16.0 limitation (see below) · AccessDeniedException → model access not enabled
    • Credential model (both-or-neither, default-chain/IRSA fallback), embedding families, image unsupported

⚠️ Finding for the broader PR (not fixed here — your call @ihoffmann-dot)

The pinned langchain4j BOM 1.0.0 resolves langchain4j-bedrock:1.0.0-beta5, which unconditionally sends stopSequences in the Converse request. Live-verified: the gpt-oss family fails with ValidationException on every chat AND streaming call, regardless of request params — while DeepSeek R1 works (and beta5 already filters its reasoning blocks correctly; no reasoning leak in either mode). Re-tested on langchain4j-bedrock:1.16.0: gpt-oss works cleanly in both modes. If gpt-oss support matters, the BOM needs a bump — left out of this stack since it affects all providers.

Checklist

  • Tests: 22/22 strategy, 38/38 factory (docs-only change)
  • No translations needed
  • No security implications

🤖 Generated with Claude Code

@sfreudenthaler @claude
docs(dotAI): document Bedrock model ID forms, credentials, and error …
d6dc436 
…mapping (#35242)

Add class-level javadoc to BedrockModelProviderStrategy covering config fields,
the both-or-neither credential model with DefaultCredentialsProvider/IRSA
fallback, inference-profile-prefixed vs bare on-demand model ID forms, common
Bedrock error meanings (on-demand-throughput -> use profile prefix; stopSequences
-> langchain4j-bedrock < 1.16.0 limitation with the gpt-oss family;
AccessDeniedException -> model access not enabled), supported embedding families,
and unsupported image generation.

Extend the ProviderConfig AWS Bedrock javadoc block to note that timeout and
maxRetries now apply to the Bedrock runtime clients, plus the model-ID-form note.

No existing dotAI provider docs markdown tree exists on this branch, so javadoc
is the documentation surface.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@github-actions
Copy link
Copy Markdown
Contributor

github-actions Bot commented Jun 5, 2026

❌ Issue Linking Required

This PR could not be linked to an issue. All PRs must be linked to an issue for tracking purposes.

How to fix this:

Option 1: Add keyword to PR body (Recommended - auto-removes this comment)
Edit this PR description and add one of these lines:

  • This PR fixes #123 or Fixes: #123

  • This PR closes #123 or Closes: #123

  • This PR resolves #123 or Resolves: #123

  • Other supported keywords: fix, fixed, close, closed, resolve, resolved
    Option 2: Link via GitHub UI (Note: won't clear the failed check)

  1. Go to the PR → Development section (right sidebar)

  2. Click "Link issue" and select an existing issue

  3. Push a new commit or re-run the workflow to clear the failed check
    Option 3: Use branch naming
    Create a new branch with one of these patterns:

  • 123-feature-description (number at start)

  • issue-123-feature-description (issue-number at start)

  • feature-issue-123 (issue-number anywhere)

Why is this required?

Issue linking ensures proper tracking, documentation, and helps maintain project history. It connects your code changes to the problem they solve.---

This comment was automatically generated by the issue linking workflow

@claude
Copy link
Copy Markdown
Contributor

claude Bot commented Jun 5, 2026 •
edited
Loading

Claude finished @sfreudenthaler's task in 50s —— View job

Rollback Safety Analysis

  • Read rollback-unsafe categories reference
  • Get PR diff
  • Analyze changes against unsafe categories
  • Apply label

Result: ✅ Safe To Rollback

This PR contains documentation-only changes — javadoc additions to two Java files:

  • BedrockModelProviderStrategy.java: New class-level Javadoc (52 lines of comments added)
  • ProviderConfig.java: Extended existing Javadoc for the Bedrock block (10 lines of comments added/modified)

No database migrations, Elasticsearch mapping changes, data model changes, API contract changes, schema alterations, or any other structural modifications were made. The changes do not touch any category in the rollback-unsafe reference (C-1 through M-4).

Label AI: Safe To Rollback has been applied.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

AI: Safe To Rollback

Projects

dotCMS - Product Planning
Status: No status

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@sfreudenthaler