assistedInstaller.py

from dataclasses import dataclass
import itertools
import time
import os
import json
from typing import Optional
import requests
from ailib import AssistedClient
import common
from logger import logger
import sys
import tenacity


@dataclass
class AssistedClientClusterInfo:
    id: str
    api_vip: str


@dataclass
class AssistedClientHostInfo:
    name: str
    status: str
    inventory: str


@dataclass
class ClusterInfo:
    name: str
    status: str


class AssistedClientAutomation(AssistedClient):  # type: ignore
    def __init__(self, url: str):
        super().__init__(url, quiet=True, debug=False)

    def cluster_exists(self, name: str) -> bool:
        return any(name == x.name for x in self.get_cluster_info_all())

    def ensure_cluster_deleted(self, name: str) -> None:
        logger.info(f"Ensuring that cluster {name} is not present")
        while self.cluster_exists(name):
            try:
                self.delete_cluster(name)
            except Exception:
                logger.info("failed to delete cluster, will retry..")
            time.sleep(5)

    def ensure_infraenv_created(self, name: str, cfg: dict[str, str]) -> None:
        if name not in (x["name"] for x in self.list_infra_envs()):
            logger.info(f"Creating infraenv {name}")
            self.create_infra_env(name, cfg)

    def ensure_infraenv_deleted(self, name: str) -> None:
        if name in (x["name"] for x in self.list_infra_envs()):
            self.delete_infra_env(name)

    @staticmethod
    def delete_kubeconfig_and_secrets(name: str, kubeconfig_path: Optional[str]) -> None:

        path, kubeconfig_path, downloaded_kubeconfig_path, downloaded_kubeadminpassword_path = common.kubeconfig_get_paths(name, kubeconfig_path)

        try:
            os.remove(kubeconfig_path)
        except OSError:
            pass

        try:
            os.remove(downloaded_kubeadminpassword_path)
        except OSError:
            pass

    def download_kubeconfig_and_secrets(
        self,
        name: str,
        kubeconfig_path: Optional[str],
        *,
        log: bool = True,
    ) -> tuple[str, str]:

        path, kubeconfig_path, downloaded_kubeconfig_path, downloaded_kubeadminpassword_path = common.kubeconfig_get_paths(name, kubeconfig_path)

        self.download_kubeconfig(name, path)
        self.download_kubeadminpassword(name, path)

        if downloaded_kubeconfig_path != kubeconfig_path:
            # download_kubeconfig() does not support specifying the full path.
            # The caller requested another name. Rename.
            os.rename(downloaded_kubeconfig_path, kubeconfig_path)

        if log:
            logger.info(f"KUBECONFIG={kubeconfig_path}")
            logger.info(f"KUBEADMIN_PASSWD={downloaded_kubeadminpassword_path}")

        return kubeconfig_path, downloaded_kubeadminpassword_path

    def download_iso_with_retry(self, infra_env: str, path: str = os.getcwd()) -> None:
        logger.info(self.info_iso(infra_env, {}))
        retries, timeout = 25, 30
        logger.info(f"Download iso from {infra_env} to {path}, retrying for {retries * timeout}s")
        for _ in range(retries):
            try:
                self.download_iso(infra_env, path)
                break
            except Exception:
                time.sleep(timeout)
        else:
            logger.error(f"Failed to download the ISO after {retries} attempts")

    def wait_cluster_ready(self, cluster_name: str) -> None:
        logger.info("Waiting for cluster state to be ready")
        cur_state = None
        while True:
            new_state = self.cluster_state(cluster_name)
            if new_state != cur_state:
                logger.info(f"Cluster state changed to {new_state}")
            cur_state = new_state
            if cur_state == "ready":
                break
            time.sleep(10)

    @tenacity.retry(wait=tenacity.wait_fixed(2), stop=tenacity.stop_after_attempt(5))
    def get_cluster_info_all(self) -> list[ClusterInfo]:
        all_clusters = self.list_clusters()
        if not isinstance(all_clusters, list):
            raise Exception(f"Unexpected type for list_clusters: {type(all_clusters)}")

        ret: list[ClusterInfo] = []
        for ci in all_clusters:
            if not isinstance(ci, dict):
                raise Exception(f"Unexpected type for list_clusters: {type(ci)}")
            if "name" not in ci or not isinstance(ci["name"], str):
                raise Exception("Invalid cluster info, no name")
            if "status" not in ci or not isinstance(ci["status"], str):
                raise Exception("Invalid cluster info, no status")
            ret.append(ClusterInfo(name=ci["name"], status=ci["status"]))
        return ret

    def cluster_state(self, cluster_name: str) -> str:
        matching_clusters = [x for x in self.get_cluster_info_all() if x.name == cluster_name]
        if len(matching_clusters) == 0:
            logger.error(f"Requested status of cluster '{cluster_name}' but couldn't find it")
            sys.exit(-1)
        elif len(matching_clusters) > 1:
            logger.error(f"Unexpected number of matching clusters: {matching_clusters}")
            sys.exit(-1)
        else:
            return matching_clusters[0].status

    def ensure_cluster_installing(self, cluster_name: str) -> None:
        self.wait_cluster_ready(cluster_name)
        self._start_until_success(cluster_name)

    def _start_until_success(self, cluster_name: str) -> None:
        logger.info(f"Starting cluster {cluster_name} (will retry until success)")
        # https://github.com/openshift/assisted-service/blob/master/swagger.yaml#L5224
        prev_cs = ""

        for tries in itertools.count(0):
            cs = self.cluster_state(cluster_name)
            if cs != prev_cs:
                logger.info(f"Cluster state is '{cs}'")
                prev_cs = cs
            if cs == "ready" or cs == "error":
                try:
                    self.start_cluster(cluster_name)
                except Exception:
                    pass
            elif cs == "installing":
                break
            time.sleep(10)
        logger.info(f"Took {tries} tries to start cluster {cluster_name}")

    def list_ai_hosts(self) -> list[AssistedClientHostInfo]:
        ret = []
        for h in filter(lambda x: "inventory" in x, self.list_hosts()):
            ret.append(AssistedClientHostInfo(h["requested_hostname"], h["status"], h["inventory"]))
        return ret

    def get_ai_host(self, name: str) -> Optional[AssistedClientHostInfo]:
        for h in self.list_ai_hosts():
            if h.name == name:
                return h
        return None

    def get_ai_ip(self, name: str, ip_range: tuple[str, str]) -> Optional[str]:
        ai_host = self.get_ai_host(name)
        if ai_host:
            inventory = json.loads(ai_host.inventory)
            routes = inventory["routes"]

            default_nics = [x['interface'] for x in routes if x['destination'] == '0.0.0.0']
            for default_nic in default_nics:
                nic_info = next(nic for nic in inventory.get('interfaces') if nic["name"] == default_nic)
                addr = str(nic_info['ipv4_addresses'][0].split('/')[0])
                if common.ip_range_contains(ip_range, addr):
                    return addr
        return None

    def allow_add_workers(self, cluster_name: str) -> None:
        uuid = self.get_ai_cluster_info(cluster_name).id
        requests.post(f"http://{self.url}/api/assisted-install/v2/clusters/{uuid}/actions/allow-add-workers")

    def get_ai_cluster_info(self, cluster_name: str) -> AssistedClientClusterInfo:
        cluster_info = self.info_cluster(cluster_name)
        if not hasattr(cluster_info, "id"):
            logger.error(f"ID is missing in cluster info for cluster {cluster_name}")
            sys.exit(-1)
        if not hasattr(cluster_info, "api_vips"):
            logger.error(f"Missing api_vips in cluster info for cluster {cluster_name}")
            sys.exit(-1)

        if len(cluster_info.api_vips) == 0:
            logger.error(f"Missing api vip in cluster info for cluster {cluster_name}")
            sys.exit(-1)

        return AssistedClientClusterInfo(cluster_info.id, cluster_info.api_vips[0].ip)