devops/shuttles/setup-cluster.ts

378 lines
13 KiB
TypeScript
Executable File

#!/usr/bin/env -S deno run --allow-run --allow-read --allow-write
// Note: TypeScript errors related to Deno imports and namespace can be safely ignored
// These are only relevant when running the script with the Deno runtime
import { Command } from "https://deno.land/x/cliffy@v1.0.0-rc.4/command/mod.ts";
import { delay } from "https://deno.land/std/async/mod.ts";
import { exists } from "https://deno.land/std/fs/mod.ts";
// Configuration constants
const alpineImage = "alpine/edge/cloud";
const alpineConfig = ['--profile', 'cloud-init-alpine'];
const archImage = "archlinux/current/cloud";
const archConfig = ['--profile', 'cloud-init-arch'];
const getIp = (i: number) => `10.110.36.${109 + i}`;
const image = archImage;
const config = archConfig;
// Enhanced logging function with timestamps and log levels
const log = {
debug: (message: string) => console.log(`[${new Date().toISOString()}] [DEBUG] ${message}`),
info: (message: string) => console.log(`[${new Date().toISOString()}] [INFO] ${message}`),
success: (message: string) => console.log(`[${new Date().toISOString()}] [SUCCESS] ✅ ${message}`),
warning: (message: string) => console.log(`[${new Date().toISOString()}] [WARNING] ⚠️ ${message}`),
error: (message: string) => console.error(`[${new Date().toISOString()}] [ERROR] ❌ ${message}`),
skip: (message: string) => console.log(`[${new Date().toISOString()}] [SKIP] ⏩ ${message}`),
};
// Helper function to execute commands with proper error handling
async function executeCommand(
cmdArray: string[],
description: string,
options: {
stdout?: "piped" | "inherit" | "null",
stderr?: "piped" | "inherit" | "null",
throwOnError?: boolean
} = {}
): Promise<{ success: boolean; output?: string; error?: string }> {
const { stdout = "piped", stderr = "piped", throwOnError = true } = options;
log.debug(`Executing: ${cmdArray.join(" ")}`);
try {
// Use Deno.Command API which is the modern replacement for Deno.run
const command = new Deno.Command(cmdArray[0], {
args: cmdArray.slice(1),
stdout: stdout === "piped" ? "piped" : stdout === "inherit" ? "inherit" : "null",
stderr: stderr === "piped" ? "piped" : stderr === "inherit" ? "inherit" : "null",
});
const { code, stdout: stdoutOutput, stderr: stderrOutput } = await command.output();
const stdoutText = stdout === "piped" ? new TextDecoder().decode(stdoutOutput).trim() : "";
const stderrText = stderr === "piped" ? new TextDecoder().decode(stderrOutput).trim() : "";
if (code !== 0) {
log.error(`Failed to ${description}: ${stderrText || "Unknown error"}`);
if (throwOnError) {
throw new Error(`Command failed: ${cmdArray.join(" ")}\n${stderrText}`);
}
}
return {
success: code === 0,
output: stdoutText,
error: stderrText
};
} catch (error) {
const errorMessage = error instanceof Error ? error.message : String(error);
log.error(`Exception while ${description}: ${errorMessage}`);
if (throwOnError) {
throw error;
}
return { success: false, error: errorMessage };
}
}
// Check if VM is ready for SSH connections
async function isVmReadyForSsh(ip: string, user: string, maxAttempts = 30): Promise<boolean> {
log.info(`Checking if VM at ${ip} is ready for SSH connections...`);
for (let attempt = 1; attempt <= maxAttempts; attempt++) {
log.debug(`SSH readiness check attempt ${attempt}/${maxAttempts}`);
const { success } = await executeCommand(
["ssh", "-o", "StrictHostKeyChecking=no", "-o", "ConnectTimeout=5", `${user}@${ip}`, "echo", "ready"],
`check SSH connectivity to ${ip}`,
{ throwOnError: false, stderr: "null" }
);
if (success) {
log.success(`VM at ${ip} is ready for SSH connections`);
return true;
}
log.debug(`VM at ${ip} not ready yet, waiting...`);
await delay(2000); // Wait 2 seconds between attempts
}
log.error(`VM at ${ip} is not ready for SSH connections after ${maxAttempts} attempts`);
return false;
}
// Check if VM is running
async function isVmRunning(vmName: string): Promise<boolean> {
const { success, output } = await executeCommand(
["incus", "list", vmName, "--format", "json"],
`check if VM ${vmName} is running`,
{ throwOnError: false }
);
if (!success || !output) {
return false;
}
try {
const vmInfo = JSON.parse(output);
return vmInfo.length > 0 && vmInfo[0].status === "Running";
} catch (error) {
const errorMessage = error instanceof Error ? error.message : String(error);
log.error(`Failed to parse VM status: ${errorMessage}`);
return false;
}
}
// Cleanup function to handle failures
async function cleanup(vmNames: string[], shouldRemove = false): Promise<void> {
log.info("Starting cleanup process...");
for (const vmName of vmNames) {
// Check if VM exists
const { success, output } = await executeCommand(
["incus", "list", vmName, "--format", "csv"],
`check if VM ${vmName} exists`,
{ throwOnError: false }
);
if (success && output) {
// Stop VM if it's running
const isRunning = await isVmRunning(vmName);
if (isRunning) {
log.info(`Stopping VM ${vmName}...`);
await executeCommand(
["incus", "stop", vmName, "--force"],
`stop VM ${vmName}`,
{ throwOnError: false }
);
}
// Remove VM if requested
if (shouldRemove) {
log.info(`Removing VM ${vmName}...`);
await executeCommand(
["incus", "delete", vmName],
`remove VM ${vmName}`,
{ throwOnError: false }
);
}
}
}
log.success("Cleanup completed");
}
const setupCluster = async (numMasters: number, forceCleanup = false) => {
log.info(`Starting setup of k3s cluster with ${numMasters} master nodes`);
const createdVMs: string[] = [];
try {
// Get hostname and user
const { output: hostname } = await executeCommand(
["hostnamectl", "hostname"],
"get hostname"
);
const { output: user } = await executeCommand(
["whoami"],
"get current user"
);
const sshKeyPubFileName = `/home/${user}/.ssh/nl.fourlights.${hostname}.pub`;
const sshKeyPrivateFileName = `/home/${user}/.ssh/nl.fourlights.${hostname}`;
// Check if SSH keys exist
if (!await exists(sshKeyPubFileName) || !await exists(sshKeyPrivateFileName)) {
log.error(`Required SSH keys not found: ${sshKeyPubFileName} or ${sshKeyPrivateFileName}`);
throw new Error("SSH keys not found");
}
// Step 1: Create Low-Resource Profile (if not exists)
const { success: profileExists } = await executeCommand(
["incus", "profile", "show", "low-resource"],
"check if low-resource profile exists",
{ stdout: "null", stderr: "null", throwOnError: false }
);
if (!profileExists) {
log.info("Creating low-resource profile...");
await executeCommand(
["incus", "profile", "create", "low-resource"],
"create low-resource profile"
);
await executeCommand(
["incus", "profile", "set", "low-resource", "limits.cpu=1", "limits.memory=512MB"],
"set low-resource profile limits"
);
await executeCommand(
["incus", "profile", "device", "add", "low-resource", "root", "disk", "pool=default", "path=/"],
"add root disk to low-resource profile"
);
// await executeCommand(
// ["incus", "profile", "device", "add", "low-resource", "eth-0", "nic", "network=incusbr0"],
// "add network interface to low-resource profile"
// );
log.success("Low-resource profile created");
} else {
log.skip("Low-resource profile already exists");
}
// Read SSH key
const sshKey = await Deno.readTextFile(sshKeyPubFileName);
// Step 3: Launch VMs (if not already running)
for (let i = 1; i <= numMasters; i++) {
const vmName = `k3s-master${i}`;
const { success: vmExists, output: vmOutput } = await executeCommand(
["incus", "list", vmName, "--format", "csv"],
`check if VM ${vmName} exists`,
{ throwOnError: false }
);
if (!vmExists || !vmOutput) {
log.info(`Creating VM ${vmName}...`);
await executeCommand(
["incus", "init", `images:${image}`, vmName, "--profile", "low-resource", "-c", "user.timezone=\"Europe/Amsterdam\"", "-c", `user.ssh_key=\"${sshKey}\"`, ...config],
`initialize VM ${vmName}`
);
await executeCommand(
["incus", "config", 'device', 'add', vmName, 'eth0', 'nic', 'nictype=bridged', 'parent=incusbr0', `ipv4.address=${getIp(i)}`],
`configure network for VM ${vmName}`
);
await executeCommand(
["incus", "start", vmName],
`start VM ${vmName}`
);
createdVMs.push(vmName);
log.success(`VM ${vmName} started`);
} else {
// Check if VM is running, if not, start it
const isRunning = await isVmRunning(vmName);
if (!isRunning) {
log.info(`Starting existing VM ${vmName}...`);
await executeCommand(
["incus", "start", vmName],
`start VM ${vmName}`
);
}
log.skip(`VM ${vmName} already exists`);
}
}
// Step 4: Install k3sup (if not installed)
const { success: k3supInstalled } = await executeCommand(
["which", "k3sup"],
"check if k3sup is installed",
{ stdout: "null", stderr: "null", throwOnError: false }
);
if (!k3supInstalled) {
log.info("Installing k3sup...");
await executeCommand(
["sh", "-c", "curl -sLS https://get.k3sup.dev | sh"],
"install k3sup"
);
log.success("k3sup installed");
} else {
log.skip("k3sup already installed");
}
// Step 5: Wait for VMs to be ready
const firstMasterIP = getIp(1);
log.info(`Waiting for first master node (${firstMasterIP}) to be ready...`);
const vmReady = await isVmReadyForSsh(firstMasterIP, "picard");
if (!vmReady) {
throw new Error(`First master node at ${firstMasterIP} is not ready for SSH connections`);
}
// Check if kubeconfig exists
const kubeconfigExists = await exists("./kubeconfig");
if (!kubeconfigExists) {
log.info("Bootstrapping first master node...");
await executeCommand(
["k3sup", "install", "--ip", firstMasterIP, "--user", "picard", "--cluster", "--ssh-key", sshKeyPrivateFileName],
"bootstrap first master node"
);
log.success("First master node bootstrapped");
} else {
log.skip("First master node already bootstrapped");
}
// Step 6: Join Additional Master Nodes (if not already joined)
for (let i = 2; i <= numMasters; i++) {
const vmName = `k3s-master${i}`;
const vmIP = getIp(i);
// Wait for VM to be ready
log.info(`Waiting for ${vmName} (${vmIP}) to be ready...`);
const nodeReady = await isVmReadyForSsh(vmIP, "picard");
if (!nodeReady) {
log.warning(`VM ${vmName} is not ready for SSH connections, skipping join operation`);
continue;
}
const { success: joined } = await executeCommand(
["kubectl", "--kubeconfig=./kubeconfig", "get", "nodes", vmName],
`check if ${vmName} has joined the cluster`,
{ stdout: "null", stderr: "null", throwOnError: false }
);
if (!joined) {
log.info(`Joining ${vmName} to the cluster...`);
await executeCommand(
["k3sup", "join", "--server", "--ip", vmIP, "--server-ip", firstMasterIP, "--user", "picard", "--ssh-key", sshKeyPrivateFileName],
`join ${vmName} to the cluster`
);
log.success(`VM ${vmName} joined the cluster`);
} else {
log.skip(`VM ${vmName} already joined the cluster`);
}
}
log.success("HA k3s cluster setup complete! 🚀");
// Verify cluster status
log.info("Verifying cluster status...");
const { success: clusterVerified, output: nodesOutput } = await executeCommand(
["kubectl", "--kubeconfig=./kubeconfig", "get", "nodes", "-o", "wide"],
"verify cluster nodes",
{ throwOnError: false }
);
if (clusterVerified) {
log.info("Cluster nodes:");
console.log(nodesOutput);
} else {
log.warning("Could not verify cluster status");
}
} catch (error) {
const errorMessage = error instanceof Error ? error.message : String(error);
log.error(`Failed to set up cluster: ${errorMessage}`);
if (createdVMs.length > 0) {
log.warning("An error occurred during setup. Cleaning up created resources...");
await cleanup(createdVMs, forceCleanup);
}
Deno.exit(1);
}
};
await new Command()
.name("setup-k3s-cluster")
.version("0.1.0")
.description("Automate the setup of an HA k3s cluster using incus and k3sup")
.option("-m, --masters <numMasters:number>", "Number of master nodes", { default: 3 })
.option("-c, --cleanup", "Force cleanup of VMs if setup fails", { default: false })
.action(({ masters, cleanup }) => setupCluster(masters, cleanup))
.parse(Deno.args);