What's Changed
New Contributors
[1.3.45] — 2026-05-20
Fixed
- CloudWatch Logs
GetLogEvents/FilterLogEventsacceptlogGroupIdentifier— both ops now resolve the target log group from the AWS-documentedlogGroupIdentifierparameter (either the bare group name or a fullarn:aws:logs:<region>:<account>:log-group:<name>[:*]ARN), as well as the originallogGroupName. Calls that pass an ARN — common from AWS SDK code that has the group ARN handy — no longer fail withResourceNotFoundException: The specified log group does not exist: None. Reported by @msulima. - ElastiCache
DescribeCacheClustersemits fullCacheNodeshape — the per-node XML now includesCacheNodeCreateTime(ISO8601),ParameterGroupStatus,CustomerAvailabilityZone(derived from the cluster's preferred AZ), andSourceCacheNodeId(when non-empty) in addition to the previousCacheNodeId/CacheNodeStatus/Endpoint.hashicorp/terraform-provider-awsv6.45.0 deref'sCacheNodeCreateTimewithout a nil check during read afteraws_elasticache_clusterapply, causingUnexpected nil pointer in: {CacheNodeCreateTime:<nil> …}and preventing Terraform from confirming cluster availability. Reported by @trackme-ddisley. - API Gateway v1
UpdateStageboolean fields parsed from patch strings — AWS sendspatchOperations[].valueas strings, but the v1 PATCH handler previously assigned them as-is via the generic patch path. SDK clients (e.g. Pulumi / AWS SDK Go v2) that read backtracingEnabledandcacheClusterEnabledthen failed deserialization withexpected Boolean to be of type *bool, got string instead. Those two root-stage fields are now coerced tobool("true"→True) before being applied, with an exact path match so a stage variable namedtracingEnabledis unaffected. Contributed by @duc12597.