Add "iso cluster" for DPU provisioning

Add support to deploy a cluster of type iso, which takes an arbitrary
iso file as an additional argument, and provisions a single host with
this ISO.

The ISO can be a local file, or an http url.

There are a number of assumptions made about the iso and BMC such that
this currently will only work for the DPU use-case, however in the
future this can be made more general to be hardware agnostic.

Example Config:

clusters:
- name: "iso-cluster"
  api_vip: "0.0.0.0"
  ingress_vip: "0.0.0.0"
  network_api_port: "eno12409"
  kind: "iso"
  install_iso: "http://myhost.redhat.com/rhel.iso"
  masters:
  - name: "acc-239"
    node: "foobar"
    type: "physical"
    bmc: "idrac.hostname.redhat.com"
    bmc_user: "root"
    bmc_password: "calvin"
    ip: "192.168.3.24"
    mac: "10:23:00:03:b5:42"

"network_api_port" should be an interface on the provisioning host
connected to the node in question.

"mac" should be the mac address of the port on the single host connected
to the "network_api_port"

Signed-off-by: Salvatore Daniele <[email protected]>

cda.py

@@ -15,7 +15,7 @@ def main_deploy(args: argparse.Namespace) -> None:
     cc = ClustersConfig(args.config, args.worker_range)
 
     # microshift does not use assisted installer so we don't need this check
-    if args.url == cc.cluster_ip_range[0] and not cc.kind == "microshift":
+    if args.url == cc.cluster_ip_range[0] and cc.kind == "openshift":
         ais = AssistedInstallerService(cc.version, args.url, cc.proxy, cc.noproxy)
         ais.start()
         # workaround, this will still install 4.14, but AI will think

clusterDeployer.py

@@ -25,6 +25,7 @@
 import microshift
 from extraConfigRunner import ExtraConfigRunner
 from clusterHost import ClusterHost
+import isoCluster
 
 
 def match_to_proper_version_format(version_cluster_config: str) -> str:
@@ -191,7 +192,7 @@ def deploy(self) -> None:
             else:
                 logger.info("Skipping pre configuration.")
 
-            if self._cc.kind != "microshift":
+            if self._cc.kind == "openshift":
                 if "masters" in self.steps:
                     self.teardown()
                     self.create_cluster()
@@ -211,7 +212,12 @@ def deploy(self) -> None:
                 microshift.deploy(self._cc.fullConfig["name"], self._cc.masters[0], self._cc.external_port, version)
             else:
                 logger.error_and_exit("Masters must be of length one for deploying microshift")
-
+        if self._cc.kind == "iso":
+            if len(self._cc.masters) == 1:
+                self.deploy_cluster_from_iso()
+            else:
+                logger.error("Masters must be of length one for deploying from iso")
+                sys.exit(-1)
         if "post" in self.steps:
             self._postconfig()
         else:
@@ -223,6 +229,9 @@ def _validate(self) -> None:
             if self._cc.masters[0].ip is None:
                 logger.error_and_exit("Missing ip on master")
 
+        if self._cc.kind == "iso":
+            return
+
         min_cores = 28
         cc = int(self._local_host.hostconn.run("nproc").out)
         if cc < min_cores:
@@ -599,3 +608,16 @@ def wait_for_workers(self) -> None:
                     logger.info(e)
 
             time.sleep(30)
+
+    def deploy_cluster_from_iso(self) -> None:
+        master = self._cc.masters[0]
+        if master.mac is None:
+            logger.error_and_exit(f"No MAC address provided for cluster {self._cc.name}, exiting")
+        if master.ip is None:
+            logger.error_and_exit(f"No IP address provided for cluster {self._cc.name}, exiting")
+        if master.name is None:
+            logger.error_and_exit(f"No name provided for cluster {self._cc.name}, exiting")
+        if not self._cc.network_api_port or self._cc.network_api_port == "auto":
+            logger.error_and_exit(f"Network API port with connection to {self._cc.name} must be specified, exiting")
+
+        isoCluster.IPUIsoBoot(self._cc, master, self._cc.install_iso)

clustersConfig.py

@@ -142,6 +142,7 @@ class ClustersConfig:
     postconfig: List[ExtraConfigArgs] = []
     ntp_source: str = "clock.redhat.com"
     base_dns_domain: str = "redhat.com"
+    install_iso: str = ""
 
     # All configurations that used to be supported but are not anymore.
     # Used to warn the user to change their config.
@@ -178,6 +179,8 @@ def __init__(self, yaml_path: str, worker_range: common.RangeList):
             self.version = cc["version"]
         if "kind" in cc:
             self.kind = cc["kind"]
+            if self.kind == "iso":
+                self.install_iso = cc["install_iso"]
         if "network_api_port" in cc:
             self.network_api_port = cc["network_api_port"]
         self.name = cc["name"]
@@ -363,7 +366,7 @@ def local_worker_vms(self) -> List[NodeConfig]:
         return [x for x in self.worker_vms() if x.node == "localhost"]
 
     def is_sno(self) -> bool:
-        return len(self.masters) == 1 and len(self.workers) == 0
+        return len(self.masters) == 1 and len(self.workers) == 0 and self.kind == "openshift"
 
 
 def main() -> None:

common.py

@@ -1,7 +1,12 @@
 from dataclasses import dataclass
 import ipaddress
-from typing import List, Optional, Set, Tuple, TypeVar, Iterator
+from typing import List, Optional, Set, Tuple, TypeVar, Iterator, Type
+from types import TracebackType
+import http.server
+import socket
+from multiprocessing import Process
 import host
+from logger import logger
 import json
 import os
 import glob
@@ -69,6 +74,49 @@ class IPRouteAddressEntry:
     addr_info: List[IPRouteAddressInfoEntry]
 
 
+class HttpServerManager:
+    def __init__(self, path: str, port: int = 8000):
+        self.path = path
+        self.port = port
+        self.process: Optional[Process] = None
+
+    def __enter__(self) -> 'HttpServerManager':
+        self.start_server()
+        return self
+
+    def __exit__(self, exc_type: Optional[Type[BaseException]], exc_value: Optional[BaseException], traceback: Optional[TracebackType]) -> None:
+        self.stop_server()
+
+    def start_server(self) -> None:
+        def target() -> None:
+            os.chdir(self.path)
+            server_address = ('', self.port)
+            httpd = http.server.HTTPServer(server_address, http.server.SimpleHTTPRequestHandler)
+            httpd.serve_forever()
+
+        self.port = self.find_open_port()
+        self.process = Process(target=target)
+        self.process.start()
+        logger.info(f"Http Server started on port {self.port}")
+
+    def stop_server(self) -> None:
+        if self.process:
+            self.process.terminate()
+            self.process.join()
+            logger.info("Http Server stopped")
+
+    def port_is_in_use(self, port: int) -> bool:
+        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+            return s.connect_ex(('localhost', port)) == 0
+
+    def find_open_port(self) -> int:
+        port = self.port
+        while self.port_is_in_use(port):
+            logger.debug(f"port {self.port} in use, trying port + 1")
+            port += 1
+        return port
+
+
 def ipa(host: host.Host) -> str:
     return host.run("ip -json a").out
 

dependencies.sh

@@ -23,7 +23,7 @@ gpgkey=https://pkgs.k8s.io/core:/stable:/v1.29/rpm/repodata/repomd.xml.key
 #exclude=kubelet kubeadm kubectl cri-tools kubernetes-cni
 EOF
 
-dnf install -y wget rust coreos-installer kubectl libvirt podman qemu-img qemu-kvm virt-install make git golang-bin virt-viewer osbuild-composer composer-cli cockpit-composer bash-completion firewalld lorax
+dnf install -y wget rust coreos-installer kubectl libvirt podman qemu-img qemu-kvm virt-install make git golang-bin virt-viewer osbuild-composer composer-cli cockpit-composer bash-completion firewalld lorax dhcp-server
 
 systemctl enable osbuild-composer.socket cockpit.socket --now
 

host.py

@@ -146,9 +146,8 @@ def from_bmc(ip_or_hostname: str, user: str = "root", password: str = "calvin")
 
     """
 
-    def boot_iso_redfish(self, iso_path: str) -> None:
+    def boot_iso_redfish(self, iso_path: str, retries: int = 10, retry_delay: int = 60) -> None:
         assert ":" in iso_path
-        retries = 10
         for attempt in range(retries):
             try:
                 self.boot_iso_with_retry(iso_path)
@@ -157,7 +156,7 @@ def boot_iso_redfish(self, iso_path: str) -> None:
                 if attempt == retries - 1:
                     raise e
                 else:
-                    time.sleep(60)
+                    time.sleep(retry_delay)
 
     def boot_iso_with_retry(self, iso_path: str) -> None:
         logger.info(iso_path)
@@ -251,10 +250,9 @@ def ssh_connect_looped(self, logins: List[Login]) -> None:
                     self._host = e.login()
                     return
                 except ssh_exception.AuthenticationException as e:
-                    logger.info(type(e))
-                    raise e
+                    logger.debug(type(e))
                 except Exception as e:
-                    logger.info(type(e))
+                    logger.debug(type(e))
                     time.sleep(10)
 
     def _rsa_login(self) -> Optional[KeyLogin]: