connectorerrors: ProvisionFailureDetail primitive for terminal Grant/Revoke#828
Open
arreyder wants to merge 2 commits into
Open
connectorerrors: ProvisionFailureDetail primitive for terminal Grant/Revoke#828arreyder wants to merge 2 commits into
arreyder wants to merge 2 commits into
Conversation
Contributor
General PR Review: connectorerrors: ProvisionFailureDetail primitive for terminal Grant/RevokeBlocking Issues: 0 | Suggestions: 0 | Threads Resolved: 0 Review SummaryThis PR adds a new Security IssuesNone found. Correctness IssuesNone found. SuggestionsNone. |
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
Connectors today return plain Go errors for upstream-permanent failures (target user SUSPENDED, group governed by a rule, license cap reached). Those resolve to gRPC code=Unknown by the time a remote caller sees them, and Unknown classifies as retryable — so callers loop on a failure that will never succeed without external state change. This change adds a baton-sdk primitive for that signal: - new proto ProvisionFailureDetail (reason + admin-facing detail) - pkg/connectorerrors.NewTerminalError builds a gRPC status with the right code (PermissionDenied / FailedPrecondition) and attaches the detail via status.WithDetails - pkg/connectorerrors.FailureDetail extracts the detail across fmt.Errorf %w and errors.Join wrap chains, via status.FromError The reason→code table picks codes the existing isClientErrorCode logic already treats as non-retryable, so callers that only look at status.Code classify these errors correctly without any changes; callers that want structured logging or admin-facing detail can read the proto. Refs OPS-1551. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Trim package doc; drop historical narrative - Add cross-reference on ResourceProvisionerV2 so the recommended interface points at the helper too - Fix IsTerminal docstring (FailureDetail allocates via FromError) - Panic on the unreachable WithDetails error path so a future field change can't silently strip the detail - Skip fmt.Sprintf when len(args)==0 so a literal upstream message containing % isn't reinterpreted - Document the Reason-is-authoritative contract on NewTerminalError / codeForReason; note forward-compat on FailureDetail - Sharpen LICENSE_EXHAUSTED (capacity) vs OPERATION_NOT_PERMITTED (policy) boundary in proto comments Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
arreyder
force-pushed
the
chris/ops-1551-provision-failure-detail
branch
from
b33ecaf to
666a4a8
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Connectors today return plain Go errors for upstream-permanent failures (target user SUSPENDED, group governed by a rule, license cap reached). Those resolve to gRPC
code=Unknownby the time a remote caller sees them, and Unknown classifies as retryable — so callers loop on a failure that will never succeed without external state change. We hit this in production with baton-okta returning E0000038 / E0000001 errors that caused ~20-minute action delays per grant.This adds the baton-sdk primitive callers and connector authors agreed was the right place for that signal (see internal review thread).
ProvisionFailureDetailwith a reason enum (USER_NOT_PROVISIONABLE,TARGET_MANAGED_EXTERNALLY,LICENSE_EXHAUSTED,VALIDATION_FAILURE,OPERATION_NOT_PERMITTED) and a free-formdetailstring for logs / admin context.pkg/connectorerrors.NewTerminalError(reason, format, args...)builds a gRPC status with the right code (PermissionDeniedforOPERATION_NOT_PERMITTED,FailedPreconditionfor everything else) and attaches the detail viastatus.WithDetails.pkg/connectorerrors.FailureDetail(err)extracts the detail acrossfmt.Errorf%wanderrors.Joinwrap chains (viastatus.FromError, which useserrors.As).ResourceProvisionerpointing connector authors at the helper.The reason→code table picks codes the existing
isClientErrorCodelogic in c1 already treats as non-retryable, so callers that only look atstatus.Code(err)classify these errors correctly without any changes; callers that want structured logging or admin-facing detail can read the proto viaFailureDetail.Transient failures (5xx, network blips, rate limits) should continue to be returned as ordinary errors — those are retryable, and rate limits have their own annotation (
RateLimitDescription).Refs OPS-1551.
Follow-ups (separate PRs)
REASON_USER_NOT_PROVISIONABLE, E0000001 (groupMembers) →REASON_TARGET_MANAGED_EXTERNALLY.FailureDetailinpkg/temporal/activity/connector-actionsfor log enrichment (reason tag interminal_marker_reason).Test plan
make protogenregenerates Go code from new protogo test ./pkg/connectorerrors/...passesmake lintclean