I saw this issue as well. The RPC API call used when we poll stack metadata is describe_stack_resource, which is overkill for what we need and returns a bunch of extra information that is thrown away by the ReST API, which was a reasonable trade-off at the time.
I suspect that it is this, and not any custom constraints, that is the source of the problem (the log message above says "Instance (...) not found", and the only log message that says "Instance" is in server_to_ipaddress(), which is only called when attempting to read the first_address attribute). However, it's entirely possible that custom constraints could also cause similar log messages.
I saw this issue as well. The RPC API call used when we poll stack metadata is describe_ stack_resource, which is overkill for what we need and returns a bunch of extra information that is thrown away by the ReST API, which was a reasonable trade-off at the time.
However, thanks to http:// git.openstack. org/cgit/ openstack/ heat/commit/ ?id=6d8a5cb35ce c81b23a71e8c6fa 692f8e7b4dca1c amongst that information we now return all the attributes of the resource, which means that attempting to read the first_address attribute is inevitable, and hence we get these warnings in the log.
I suspect that it is this, and not any custom constraints, that is the source of the problem (the log message above says "Instance (...) not found", and the only log message that says "Instance" is in server_ to_ipaddress( ), which is only called when attempting to read the first_address attribute). However, it's entirely possible that custom constraints could also cause similar log messages.