There must be a more idiomatic way

sample.py

        for container in containers:
            p1 = re.compile('.*?kinto_read-only_(\d)$')
            p2 = re.compile('.*?kinto_master_(\d)$')
            print container['Names']
            if container['Names']:
                for name in container['Names']:
                    if p1.match(name):
                        ports = container['Ports']
                        for port in ports:
                            self.read_only['ip'] = port['IP']
                            self.read_only['port'] = port['PrivatePort']
                    if p2.match(name):
                        ports = container['Ports']
                        for port in ports:
                            self.master['ip'] = port['IP']
                            self.master['port'] = port['PrivatePort']

_RE_READ_ONLY = re.compile('.*?kinto_read-only_(\d)$')
_RE_MASTER = re.compile('.*?kinto_master_(\d)$')

def get_ports(self, containers):
    # use a list comprehension to filter out containers with names
    containers = [container for container in containers if 'Names' in container]
    for container in containers:
        if has_match(_RE_READ_ONLY, container['Names']):
            # seems like you are overriding, do you only want the last one?
            for port in container.get('Ports', {}):
                self.read_only['ip'] = port['IP']
                self.read_only['port'] = port['PrivatePort']

        if has_match(_RE_MASTER, container['Names']):
            for port in container.get('Ports', {}):
                self.master['ip'] = port['IP']
                self.master['port'] = port['PrivatePort']


def has_match(r, names):
    # check using a list comprehension
    return true in [true for name in names if r.match(name)]

nitpick: I would not use list comprehensions to filter out things, because it means you are looping over the whole list at most twice -- which is suboptimal