bug: Route53 weighted alias cannot be recreated after successful DELETE (dev102)

[AndreKurait]

A weighted Route53 alias cannot be recreated after a successful DELETE on the current LocalStack Pro dev image. DELETE returns HTTP 200 with INSYNC, and ListResourceRecordSets confirms the exact name/type/SetIdentifier is absent. Recreating the original record nevertheless returns InvalidChangeBatch: already exists.

Reproduced on 2026-10-12 with the unmodified vendor image, its default entrypoint, SERVICES=route53, and no custom extensions/init scripts. The reproduction uses one fresh hosted zone, two ordinary A records as same-zone alias targets, and fictional account 111111111111. No ELB, CloudFormation or application harness is involved.

Environment:

• localstack/localstack-pro:dev, response header x-localstack: 2026.10.0.dev102
• Pulled multiarchitecture index: sha256:c9929d6cb8e737b3192348812b1b92f5c6495732d8f6db1e921547be5673479b
• Actual image config ID: sha256:a14e9b1127cbb40029bf942a06114cb2fa77bb2b7d038fe00a469bb27c3e6b87
• Linux ARM64, bundled Python/boto3; x86 has not been independently tested for this reproduction.

Observed sequence:

1. CREATE two ordinary A records: HTTP 200 / INSYNC.
2. CREATE a weighted A alias pointing to the first record: HTTP 200 / INSYNC.
3. UPSERT that same name/type/SetIdentifier to the second target: HTTP 200 / INSYNC.
4. DELETE the exact updated record: HTTP 200 / INSYNC.
5. ListResourceRecordSets shows the weighted alias is absent.
6. CREATE the original weighted alias: HTTP 400 / InvalidChangeBatch, claiming it already exists.

Expected: step 6 succeeds after successful deletion. The same tested script subsequently deleted the remaining records and hosted zone without cleanup errors.

The actual error was:

Tried to create resource record set [name='customer.weighted-alias-63afd9932f66.fixture.invalid.', type='A'] but it already exists

This resembles the previously closed localstack/localstack#12910, but this report supplies a current dev-image weighted-alias reproduction; I have not established that the underlying cause is the same.

Save the following exact tested script as repro.py. A minimal startup command for the same pinned image is below; LOCALSTACK_AUTH_TOKEN should already be set through your normal license workflow. Wait for the Route53 service to be available in /_localstack/health before running the script.

docker run --rm -d --name route53-repro --platform linux/arm64 \
  -e LOCALSTACK_AUTH_TOKEN -e SERVICES=route53 -e DNS_ADDRESS=0 \
  localstack/localstack-pro@sha256:c9929d6cb8e737b3192348812b1b92f5c6495732d8f6db1e921547be5673479b

Execute using the bundled SDK (no host port publication is needed):

docker exec -i \
  -e FIXTURE_ENDPOINT=http://127.0.0.1:4566 \
  -e FIXTURE_ENDPOINT_HOST=127.0.0.1 \
  route53-repro /opt/code/localstack/.venv/bin/python - < repro.py

The script prints each request, response and public record listing, and cleans up its uniquely named hosted zone even after the expected reproduction failure.

import copy, json, os, sys, uuid
from urllib.parse import urlparse
import boto3
from botocore.config import Config

def main():
 endpoint=os.environ["FIXTURE_ENDPOINT"]
 parsed=urlparse(endpoint)
 assert parsed.scheme=="http" and parsed.hostname and not parsed.username and not parsed.password and parsed.path in ("", "/")
 assert parsed.hostname==os.environ["FIXTURE_ENDPOINT_HOST"], "Endpoint must match the outer fixture admitted host"
 client=boto3.client("route53", endpoint_url=endpoint, region_name="eu-west-1", aws_access_key_id="111111111111", aws_secret_access_key="fixture", config=Config(connect_timeout=3, read_timeout=10, retries={"max_attempts":0}))
 name="weighted-alias-"+uuid.uuid4().hex[:12]+".fixture.invalid"
 result={"status":"FAILED", "scope":"Vanilla vendor public Route53 weighted alias mutation contract; no harness extensions or HTTP target requests", "zone_name":name, "events":[], "cleanup_errors":[]}
 zone=None
 def emit(label, **values):
  row={"step":label, **values};result["events"].append(row);print(json.dumps({"event":"PUBLIC_ROUTE53_STATE", **row}, default=str), flush=True)
 def change(label, action, record):
  request={"HostedZoneId":zone, "ChangeBatch":{"Changes":[{"Action":action,"ResourceRecordSet":copy.deepcopy(record)}]}}
  try:response=client.change_resource_record_sets(**request)
  except Exception as error:
   emit(label,request=request,error_type=type(error).__name__,error_response=getattr(error,"response",None));raise
  emit(label,request=request,response=response)
  assert response["ResponseMetadata"]["HTTPStatusCode"]==200
 def listing(label):
  response=client.list_resource_record_sets(HostedZoneId=zone);emit(label,response=response);assert not response.get("IsTruncated");return response["ResourceRecordSets"]
 try:
  created=client.create_hosted_zone(Name=name, CallerReference=name);zone=created["HostedZone"]["Id"];emit("create-zone",response=created)
  for suffix,ip in (("first","192.0.2.10"),("second","192.0.2.11")):
   change("create-target-"+suffix,"CREATE",{"Name":suffix+"."+name,"Type":"A","TTL":60,"ResourceRecords":[{"Value":ip}]})
  record={"Name":"customer."+name,"Type":"A","SetIdentifier":"owned-weighted","Weight":1,"AliasTarget":{"HostedZoneId":zone.split("/")[-1],"DNSName":"first."+name,"EvaluateTargetHealth":False}}
  change("create-weighted","CREATE",record);listing("after-create")
  repointed=copy.deepcopy(record);repointed["AliasTarget"]["DNSName"]="second."+name
  change("upsert-weighted","UPSERT",repointed);listing("after-upsert")
  change("delete-weighted","DELETE",repointed);remaining=listing("after-delete")
  assert not any(r["Name"].rstrip(".")==record["Name"] and r["Type"]=="A" and r.get("SetIdentifier")==record["SetIdentifier"] for r in remaining)
  change("recreate-weighted","CREATE",record);listing("after-recreate");result["status"]="PASS"
 except Exception as error:
  result["error"]={"type":type(error).__name__,"message":str(error),"response":getattr(error,"response",None)}
 finally:
  if zone:
   try:
    for record in client.list_resource_record_sets(HostedZoneId=zone)["ResourceRecordSets"]:
     if record["Type"] not in ("NS","SOA"):
      client.change_resource_record_sets(HostedZoneId=zone,ChangeBatch={"Changes":[{"Action":"DELETE","ResourceRecordSet":record}]})
    client.delete_hosted_zone(Id=zone)
   except Exception as error:result["cleanup_errors"].append({"type":type(error).__name__,"message":str(error)})
  print(json.dumps({"event":"VANILLA_ROUTE53_RESULT",**result},default=str),flush=True)
 return 0 if result["status"]=="PASS" and not result["cleanup_errors"] else 1
if __name__=="__main__":sys.exit(main())

[EddEcho]

Hi @AndreKurait , thanks for the report. This is now fixed in our latest dev build. Could you run docker pull localstack/localstack-pro:dev, give it a try, and let us know if it works for you