feat: add various infrastructure check and automation scripts

This commit is contained in:
2026-11-15 17:17:41 -05:00
parent 336b7759d0
commit b301aa3ece
232 changed files with 8274 additions and 298 deletions

View File

@@ -0,0 +1,146 @@
import paramiko
import json
SSH_KEY = r"C:\Users\Snarf\.ssh\id_ed25519_snarfnet"
nodes = {
"snaisvr44hrvstr01": "140.44.4.81",
"snaisvr44hrvstr02": "140.44.4.82",
"snaisvr44hrvstr03": "140.44.4.83",
"snaisvr44hrvstr04": "140.44.4.84"
}
# Python script that runs on each node
node_script = """python3 - << 'PYEOF'
import os
import subprocess
import json
import glob
# Clean up any bad units first
for f in glob.glob("/etc/systemd/system/var-lib-harvester-extra*"):
if os.path.isfile(f) or os.path.islink(f):
os.remove(f)
# 1. Map physical disks to their Longhorn diskUUID
out = subprocess.check_output("blkid", shell=True).decode()
disks = {}
for line in out.splitlines():
if not line.startswith("/dev/sd"): continue
dev = line.split(":")[0]
if "sda" in dev: continue
uuid = ""
for part in line.split():
if part.startswith("UUID="):
uuid = part.split('"')[1]
if uuid:
disks[dev] = uuid
disk_info = {} # diskUUID -> fs_uuid
for dev, uuid in disks.items():
tmp = f"/tmp/m_{os.path.basename(dev)}"
os.makedirs(tmp, exist_ok=True)
try:
subprocess.check_call(f"mount {dev} {tmp}", shell=True)
cfg_file = os.path.join(tmp, "longhorn-disk.cfg")
if os.path.exists(cfg_file):
with open(cfg_file) as f:
data = json.load(f)
lh_uuid = data.get("diskUUID")
disk_info[lh_uuid] = uuid
subprocess.call(f"umount {tmp}", shell=True)
except Exception as e:
pass
# 2. Read expected from /tmp/node_expected.json
with open("/tmp/node_expected.json") as f:
expected = json.load(f)
created_units = []
for disk_name, info in expected.items():
lh_uuid = info["diskUUID"]
path = info["path"]
if not path or path == "/var/lib/harvester/defaultdisk":
continue
if lh_uuid in disk_info:
fs_uuid = disk_info[lh_uuid]
os.makedirs(path, exist_ok=True)
# Use systemd-escape binary to get the exact canonical unit filename
unit_name = subprocess.check_output(["systemd-escape", "-p", "--suffix=mount", path]).decode().strip()
unit_path = os.path.join("/etc/systemd/system", unit_name)
unit_content = f\"\"\"[Unit]
Description=Harvester Extra Disk {disk_name}
Before=rke2-server.service rke2-agent.service
[Mount]
What=/dev/disk/by-uuid/{fs_uuid}
Where={path}
Type=ext4
Options=defaults
[Install]
WantedBy=multi-user.target
\"\"\"
with open(unit_path, "w") as uf:
uf.write(unit_content)
created_units.append(unit_name)
# Ensure it is mounted right now
mounts = subprocess.check_output("mount", shell=True).decode()
if path not in mounts:
subprocess.call(f"mount /dev/disk/by-uuid/{fs_uuid} {path}", shell=True)
subprocess.call(["systemctl", "daemon-reload"])
for u in created_units:
subprocess.call(["systemctl", "enable", u])
subprocess.call(["systemctl", "start", u])
print("Configured & mounted units:", created_units)
PYEOF
"""
# Fetch latest expected disks from Harvester VIP
c = paramiko.SSHClient()
c.set_missing_host_key_policy(paramiko.AutoAddPolicy())
c.connect("140.44.4.90", username="rancher", key_filename=SSH_KEY, timeout=10)
_, o, _ = c.exec_command("echo Katzir476! | sudo -S sh -c 'export KUBECONFIG=/etc/rancher/rke2/rke2.yaml; PATH=$PATH:/var/lib/rancher/rke2/bin; kubectl get nodes.longhorn.io -n longhorn-system -o json'")
data = json.loads(o.read().decode())
c.close()
all_expected = {}
for node in data.get("items", []):
nname = node["metadata"]["name"]
all_expected[nname] = {}
for dname, dstat in node.get("status", {}).get("diskStatus", {}).items():
all_expected[nname][dname] = {
"path": node.get("spec", {}).get("disks", {}).get(dname, {}).get("path"),
"diskUUID": dstat.get("diskUUID")
}
for hname, ip in nodes.items():
print(f"\n=================== Deploying to {hname} ({ip}) ===================")
expected_for_node = all_expected.get(hname, {})
nc = paramiko.SSHClient()
nc.set_missing_host_key_policy(paramiko.AutoAddPolicy())
try:
nc.connect(ip, username="rancher", key_filename=SSH_KEY, timeout=10)
# Write /tmp/node_expected.json via stdin
stdin, stdout, stderr = nc.exec_command("cat > /tmp/node_expected.json")
stdin.write(json.dumps(expected_for_node))
stdin.close()
stdout.channel.recv_exit_status()
_, o, e = nc.exec_command(f"echo Katzir476! | sudo -S {node_script}")
print(o.read().decode())
if e.read().decode():
pass
_, o, _ = nc.exec_command("df -h | grep extra-disks")
print(f"Active mounts on {hname}:\n" + o.read().decode())
nc.close()
except Exception as e:
print(f"Error on {hname}: {e}")
print("\nAll nodes deployed successfully!")