Deploy Service does not honor space constraint when searching for service instances

[krmeda]

As seen in the below code:

python-cf-api/cf_api/deploy_service.py

Lines 122 to 124 in 477748b

	res = self._cc.request(self._space.service_instances_url)\
	.get_by_name(name)
	self._service_instance = res.resource

The understanding from the service is that the service instance is restrained to the space resource that is set already. It causes undefined behavior when there are more than one service instance with the same name in multiple spaces in the same Org (and the user has access to both).

[adamjaso]

I believe I understand what you are saying here... You are suggesting that the code you cited is incorrect, because it is looking for a specific service in a specific space, but it appears to be querying without reference to the space, in which case it may match another service of the same name in the same org. Correct? Did you find this because it has caused you problems, or did you find it only by looking at the code? From what I can see, the url for self._space.service_instances_url will look like this /v2/spaces/<SPACEGUID>/service_instances. This would make it so that the search for the service instance is actually scoped to the space, as it should be and would not match a service instance from another org or space. @keerthimr

[adamjaso]

@keerthimr Can you confirm if this is causing you trouble now?

[krmeda]

Sorry looks like i misunderstood the cause of issue.
When I try to create multiple services only the first one gets created and the result for all the other invocations are references to the first created service.

space = Space(cc, org_name=args.cfOrg, space_name=args.cfSpace)
if space.space is None:
    logger.warn("Specified space: %s does not exist", args.cfSpace)
    raise Exception("Specified Space does not exist")

serviceDefs = yaml.load(args.service_def_file)
# logger.debug("Service Definitions: %s", serviceDefs)

serviceDefs = map(deserialize_service_definition, serviceDefs['services'])

logger.debug("Service Definitions: %s", serviceDefs)
svcsToTrack = {}
service = space.get_deploy_service()
for svcDef in serviceDefs:
    logger.debug("Querying Service Info for: %s", svcDef)
    cfSvcInst = service.create(name=svcDef.instance_name, service_name=svcDef.service_name, service_plan_name=svcDef.service_plan,
               tags=svcDef.tags, parameters=svcDef.parameters)
    logger.info("Service Details: %s",cfSvcInst)
    svcsToTrack[svcDef.instance_name]= CfServiceInstance(name=svcDef.instance_name ,service_deploy=service, service_resource=cfSvcInst)

The svcsToTrack dict has the same return values for all of them

[adamjaso]

Ah. I think I see what you are meaning. Sorry for the misunderstanding. I will work on a fix for that. In the mean time, here is a workaround:

# ...
for svcDef in serviceDefs:
    service = space.get_deploy_service()
    cfSvcInst = service.create(name=svcDef.instance_name, service_name=svcDef.service_name, service_plan_name=svcDef.service_plan,
               tags=svcDef.tags, parameters=svcDef.parameters)
# ...