fix: spawn quick-wins #103 #104 #108 #110 (truffle-backed validation) - #111
Merged
Conversation
#103 FSx cross-region S3 bucket: HeadBucket answers 301/PermanentRedirect (not NotFound) for a bucket in another region. Treat that as "exists, don't create" instead of erroring. Catch both the smithy APIError code and a bare HTTP 301 ResponseError. #104 MPI placement group on HPC types: hpc6a/hpc7a/hpc7g don't support cluster placement groups (they use AWS HPC networking). --auto-placement-group (default on) now SKIPs them gracefully; only an explicit --placement-group errors when unsupported. Capability comes from truffle GetCapabilities, not a static list. #108 RunInstances idempotency + classifiable errors: add optional LaunchConfig.ClientToken (deterministic per cluster/entity/generation for cohort callers) and wrap RunInstances failures in LaunchError{Code} carrying the verbatim AWS error code, so callers classify capacity/quota/config faults on a code rather than string-matching. Empty token = today's behavior. #110 pre-flight instance-type constraints: validate EFA / hibernation / MPI support BEFORE creating IAM role or security group, with actionable errors (truffle-find hints). One truffle GetCapabilities call backs all checks; HPC types are exempt from the placement-group requirement for --mpi. spawn now consumes truffle (v0.38.1) as the instance-type capability authority rather than re-querying EC2 itself — one-way dep, no cycle. isHPCInstanceType uses the "hpc<digit>" naming contract so new HPC families are covered without a hardcoded list.
Codecov Report❌ Patch coverage is
📢 Thoughts on this report? Let us know! |
Extract the HeadBucket 301 detection into a pure isCrossRegionRedirect helper and table-test both shapes (coded APIError + bare HTTP-301 ResponseError) plus negatives. Add unit tests for newLaunchError code extraction / Unwrap chain. Closes the patch-coverage gap on the quick-wins branch.
This was referenced Jun 12, 2026
Closed
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.
Batches four spawn bug-fixes. Each is self-contained; all unit/substrate-tested,
make checkgreen.#103 — FSx cross-region S3 bucket 301
HeadBucketanswers 301 PermanentRedirect/MovedPermanently (notNotFound) for a bucket that exists in another region. We were erroring on it. Now treat 301 as "exists, don't create" — catching both the smithyAPIErrorcode and a bare HTTP-301ResponseError(the SDK surfaces it both ways). FSx itself still validates cross-region usability downstream.#104 — MPI placement group on HPC instance types
hpc6a/hpc7a/hpc7g/… don't support cluster placement groups — they get low-latency networking from AWS HPC infrastructure.--auto-placement-group(on by default) now skips them gracefully with an info line; only an explicit--placement-grouperrors when unsupported. Support is read from truffleGetCapabilities, not a static allowlist.#108 — RunInstances idempotency + classifiable errors
LaunchConfig.ClientToken→ set asRunInstancesClientToken(deterministic per cluster/entity/generation for cohort callers; empty = today's behavior, so no double-launch on a retried network timeout).RunInstancesfailures wrapped inLaunchError{Code}carrying the verbatim AWS error code (InsufficientInstanceCapacity,RequestLimitExceeded, …) so callers classify capacity/quota/config faults on a code, not a string match.Unwrappreserved, soerrors.As(&smithyAPIError)still works.#110 — pre-flight instance-type constraints
Validate EFA / hibernation / MPI support before creating the IAM role or security group, with actionable errors (
truffle findhints). One truffleGetCapabilitiescall backs all checks. HPC types are exempt from the placement-group requirement for--mpi(consistent with #104).truffle integration: spawn now consumes truffle v0.38.1 as the instance-type capability authority instead of re-querying EC2 from spawn. One-way dependency (spawn→truffle), no cycle. Required bumping truffle's substrate dep to v0.70.0 (matching spawn) so the module graph resolves under
go mod tidy— done in spore-host/truffle#49.isHPCInstanceTypeuses thehpc<digit>naming contract so new HPC families are covered without a hardcoded list.Deferred (out of scope): #106 SageMaker (code parked in ~/src/fieldwork), #107 spored notifications, #109 q0 readiness tags, #101 --generalize warm AMI.