Send periodic "running" status messages while nixos-rebuild executes, preventing the idle timeout from triggering before deployments complete. This fixes false "Some deployments failed" warnings in MCP when builds take longer than 30 seconds. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
161 lines
4.2 KiB
Go
161 lines
4.2 KiB
Go
package deploy
|
|
|
|
import (
|
|
"bytes"
|
|
"context"
|
|
"fmt"
|
|
"os/exec"
|
|
"time"
|
|
|
|
"git.t-juice.club/torjus/homelab-deploy/internal/messages"
|
|
)
|
|
|
|
// Executor handles the execution of nixos-rebuild commands.
|
|
type Executor struct {
|
|
flakeURL string
|
|
hostname string
|
|
timeout time.Duration
|
|
}
|
|
|
|
// NewExecutor creates a new deployment executor.
|
|
func NewExecutor(flakeURL, hostname string, timeout time.Duration) *Executor {
|
|
return &Executor{
|
|
flakeURL: flakeURL,
|
|
hostname: hostname,
|
|
timeout: timeout,
|
|
}
|
|
}
|
|
|
|
// Result contains the result of a deployment execution.
|
|
type Result struct {
|
|
Success bool
|
|
ExitCode int
|
|
Stdout string
|
|
Stderr string
|
|
Error error
|
|
}
|
|
|
|
// ExecuteOptions contains optional settings for Execute.
|
|
type ExecuteOptions struct {
|
|
// HeartbeatInterval is how often to call the heartbeat callback.
|
|
// If zero, no heartbeat is sent.
|
|
HeartbeatInterval time.Duration
|
|
// HeartbeatCallback is called periodically with elapsed time while the command runs.
|
|
HeartbeatCallback func(elapsed time.Duration)
|
|
}
|
|
|
|
// ValidateRevision checks if a revision exists in the remote repository.
|
|
// It uses git ls-remote to verify the ref exists.
|
|
func (e *Executor) ValidateRevision(ctx context.Context, revision string) error {
|
|
// Extract the base URL for git ls-remote
|
|
// flakeURL is like git+https://git.example.com/user/repo.git
|
|
// We need to strip the git+ prefix for git ls-remote
|
|
gitURL := e.flakeURL
|
|
if len(gitURL) > 4 && gitURL[:4] == "git+" {
|
|
gitURL = gitURL[4:]
|
|
}
|
|
|
|
ctx, cancel := context.WithTimeout(ctx, 30*time.Second)
|
|
defer cancel()
|
|
|
|
cmd := exec.CommandContext(ctx, "git", "ls-remote", "--exit-code", gitURL, revision)
|
|
var stderr bytes.Buffer
|
|
cmd.Stderr = &stderr
|
|
|
|
if err := cmd.Run(); err != nil {
|
|
if ctx.Err() == context.DeadlineExceeded {
|
|
return fmt.Errorf("timeout validating revision")
|
|
}
|
|
return fmt.Errorf("revision %q not found: %w", revision, err)
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
// Execute runs nixos-rebuild with the specified action and revision.
|
|
func (e *Executor) Execute(ctx context.Context, action messages.Action, revision string) *Result {
|
|
return e.ExecuteWithOptions(ctx, action, revision, nil)
|
|
}
|
|
|
|
// ExecuteWithOptions runs nixos-rebuild with the specified action, revision, and options.
|
|
func (e *Executor) ExecuteWithOptions(ctx context.Context, action messages.Action, revision string, opts *ExecuteOptions) *Result {
|
|
ctx, cancel := context.WithTimeout(ctx, e.timeout)
|
|
defer cancel()
|
|
|
|
// Build the flake reference: <flake-url>?ref=<revision>#<hostname>
|
|
flakeRef := fmt.Sprintf("%s?ref=%s#%s", e.flakeURL, revision, e.hostname)
|
|
|
|
cmd := exec.CommandContext(ctx, "nixos-rebuild", string(action), "--flake", flakeRef)
|
|
|
|
var stdout, stderr bytes.Buffer
|
|
cmd.Stdout = &stdout
|
|
cmd.Stderr = &stderr
|
|
|
|
// Start the command
|
|
startTime := time.Now()
|
|
if err := cmd.Start(); err != nil {
|
|
return &Result{
|
|
Success: false,
|
|
ExitCode: -1,
|
|
Error: fmt.Errorf("failed to start command: %w", err),
|
|
}
|
|
}
|
|
|
|
// Set up heartbeat if configured
|
|
var heartbeatDone chan struct{}
|
|
if opts != nil && opts.HeartbeatInterval > 0 && opts.HeartbeatCallback != nil {
|
|
heartbeatDone = make(chan struct{})
|
|
go func() {
|
|
ticker := time.NewTicker(opts.HeartbeatInterval)
|
|
defer ticker.Stop()
|
|
for {
|
|
select {
|
|
case <-heartbeatDone:
|
|
return
|
|
case <-ticker.C:
|
|
opts.HeartbeatCallback(time.Since(startTime))
|
|
}
|
|
}
|
|
}()
|
|
}
|
|
|
|
// Wait for command to complete
|
|
err := cmd.Wait()
|
|
|
|
// Stop heartbeat goroutine
|
|
if heartbeatDone != nil {
|
|
close(heartbeatDone)
|
|
}
|
|
|
|
result := &Result{
|
|
Stdout: stdout.String(),
|
|
Stderr: stderr.String(),
|
|
}
|
|
|
|
if err != nil {
|
|
result.Success = false
|
|
result.Error = err
|
|
|
|
if ctx.Err() == context.DeadlineExceeded {
|
|
result.Error = fmt.Errorf("deployment timed out after %v", e.timeout)
|
|
}
|
|
|
|
if exitErr, ok := err.(*exec.ExitError); ok {
|
|
result.ExitCode = exitErr.ExitCode()
|
|
} else {
|
|
result.ExitCode = -1
|
|
}
|
|
} else {
|
|
result.Success = true
|
|
result.ExitCode = 0
|
|
}
|
|
|
|
return result
|
|
}
|
|
|
|
// BuildCommand returns the command that would be executed (for logging/debugging).
|
|
func (e *Executor) BuildCommand(action messages.Action, revision string) string {
|
|
flakeRef := fmt.Sprintf("%s?ref=%s#%s", e.flakeURL, revision, e.hostname)
|
|
return fmt.Sprintf("nixos-rebuild %s --flake %s", action, flakeRef)
|
|
}
|