Files
stalwart-migrator/cmd/stalwart-migrate/rehearse.go
T
jcoffey-dev 9faa21f4f1 Fix the three defects that cost a production restore
A live migration on 2026-08-24 stopped a production mail server and then
discovered the host's stalwart-cli was 0.13.4 - present, but from when the
CLI shipped with the server, with no `apply` command. The migration needs
v1.0.2+ from the separately-versioned stalwartlabs/cli repository.

Recovery was closed in both directions. v0.16's recovery-mode boot had
already bumped the store schema to v6, so the 0.15.5 binary refused to
reopen it ("expected 5 or below, found 6"). Going forward needed
export.json, which this tool's own failure path had deleted - and
regenerating it required a settings dump from a live v0.15 instance that
could no longer start. The operator restored a day-old snapshot and lost a
day of mail across nine domains.

Three fixes:

1. preflight.CheckExternalTools verifies stalwart-cli exists and is v1.0.2
   or later, and that python3 runs - before anything is touched. Every fact
   needed to prevent this was available in under a second from a stopped
   state. Skipped for a patch upgrade, which invokes neither tool.

2. A failed run no longer deletes its work directory. Cleaning up on every
   exit path was right for a sandboxed rehearsal and catastrophic here:
   once the service is stopped the settings dump cannot be regenerated, so
   deleting it removes the only way forward. The failure now prints the
   resume command instead.

3. `run --resume <id>` continues an interrupted run. The checkpoint
   machinery existed but never engaged, because run created a new run every
   invocation - so a retry re-ran preflight against a binary already moved
   aside, and failed. Completed steps are skipped from the checkpoint.

Proven against a VM built to match the failure: stalwart-cli 0.15.5,
accounts and mail seeded.

  * preflight refused, service still active, mail still accepted
  * a stub CLI passing --version and failing apply left the run stopped
    with all eight inputs intact and the resume command printed
  * --resume carried it to a clean finish: five seconds of downtime,
    listeners regenerated, admin role restored, quotas rebuilt

That failure-path test is the one that should have run before production.
Every earlier test had stalwart-cli installed from the start, and the one
failure I did exercise happened to leave its artifacts behind.
2026-08-23 23:20:47 -07:00

335 lines
14 KiB
Go

// SPDX-FileCopyrightText: 2026 LINUXexpert-org
// SPDX-License-Identifier: GPL-3.0-or-later
package main
import (
"context"
"flag"
"fmt"
"io"
"net/http"
"os"
"path/filepath"
"github.com/LINUXexpert-org/stalwart-migrator/internal/applyplan"
"github.com/LINUXexpert-org/stalwart-migrator/internal/backup"
"github.com/LINUXexpert-org/stalwart-migrator/internal/checkpoint"
"github.com/LINUXexpert-org/stalwart-migrator/internal/plan"
"github.com/LINUXexpert-org/stalwart-migrator/internal/preflight"
)
// runRehearse implements `stalwart-migrate rehearse` (ARCHITECTURE.md §4.9):
// preflight, dump this instance's settings and principals, convert them with
// migrate_v016.py, and report both halves of the result - what will carry
// over, and what will not.
//
// It copies no data, clones nothing, starts no server, and never writes to
// the store, so it is safe to run against production repeatedly and without
// a maintenance window. That is a deliberate narrowing from the sandbox-
// cloning dry run this replaces: cloning the store proved only that the
// store migrates and opens, at the cost of copying the data twice and
// reading a live store to do it, while the half that found every real
// problem - an empty defaultHostname v0.16 rejects, passwords v0.16 refuses,
// and a 12,182-key reconstruction worklist - needs no copy at all.
//
// The worklist is the point. Measured against a real production instance,
// migrate_v016.py carried 219 of 12,401 settings; server.listener was not
// among them, so a migrated instance answers on no ports until an operator
// rebuilds them. Anything that reported such a migration as a success
// without saying so would be actively misleading.
func runRehearse(args []string) (err error) {
fs := flag.NewFlagSet("rehearse", flag.ExitOnError)
binaryPath := fs.String("binary", "/usr/local/bin/stalwart", "path to the currently-installed stalwart binary")
configPath := fs.String("config", "/etc/stalwart/config.toml", "path to stalwart's current config file")
dataDir := fs.String("data-dir", "/var/lib/stalwart", "stalwart data directory (read-only here; used by preflight's checks)")
containerName := fs.String("container", "stalwart", "docker container name, if applicable")
adminURL := fs.String("admin-url", "", "base URL for the live instance's admin/JMAP API (required)")
adminUser := fs.String("admin-user", "", "admin username")
adminPassword := fs.String("admin-password", os.Getenv("STALWART_MIGRATE_ADMIN_PASSWORD"),
"admin password (or set STALWART_MIGRATE_ADMIN_PASSWORD)")
targetVersion := fs.String("target", "latest", `target Stalwart version, or "latest"`)
stateDir := fs.String("state-dir", checkpoint.DefaultBaseDir, "directory to store run checkpoints in")
workDir := fs.String("work-dir", "/var/lib/stalwart-migrator/work", "scratch directory for the dumps and converted plan (cleaned up afterward - see --keep-artifacts)")
pythonPath := fs.String("python", "python3", "path to python3")
stalwartCLI := fs.String("stalwart-cli", "stalwart-cli",
"path to stalwart-cli (v1.0.2 or later; a separate download from the server) - rehearsal doesn't invoke it, but checks it so `run` doesn't fail after stopping the service")
migrationScriptSHA256 := fs.String("migration-script-sha256", "", "pinned sha256 of migrate_v016.py (recommended; the first run prints the hash to pin)")
minFree := fs.Float64("min-free-multiple", 2.0, "free-space multiple preflight checks for; rehearsal itself copies nothing")
keepArtifacts := fs.Bool("keep-artifacts", false, "don't delete work-dir/<run-id> afterward")
if err := fs.Parse(args); err != nil {
return err
}
if *adminURL == "" {
return fmt.Errorf("--admin-url is required: rehearsal converts the settings this instance actually has, which means reading them from it")
}
ctx := context.Background()
httpClient := &http.Client{}
if err := os.MkdirAll(*workDir, 0o750); err != nil {
return fmt.Errorf("create work dir %s: %w", *workDir, err)
}
store := checkpoint.NewStore(*stateDir)
rs, err := store.Create("", *targetVersion)
if err != nil {
return fmt.Errorf("create run: %w", err)
}
fmt.Printf("run id: %s\n\n", rs.RunID)
runWorkDir := filepath.Join(*workDir, rs.RunID)
runStateDir := filepath.Join(*stateDir, rs.RunID)
// The rehearsal's two conclusions survive cleanup: the worklist of what
// won't carry over, and the plan of what will. Everything else in the
// work directory is scratch. Recording an artifact that points into a
// directory about to be deleted would leave the checkpoint referring to
// files that aren't there.
keptWorklist := filepath.Join(runStateDir, "unmigrated.txt")
keptPlan := filepath.Join(runStateDir, "export.json")
keptSupplement := filepath.Join(runStateDir, "supplement.json")
defer func() {
if _, statErr := os.Stat(runWorkDir); os.IsNotExist(statErr) {
return
}
if *keepArtifacts {
fmt.Printf("\nartifacts kept at %s (--keep-artifacts)\n", runWorkDir)
return
}
if err != nil {
fmt.Fprintf(os.Stderr, "\nthe rehearsal failed; its artifacts are kept at %s for inspection\n", runWorkDir)
return
}
if rmErr := os.RemoveAll(runWorkDir); rmErr != nil {
fmt.Fprintf(os.Stderr, "\nwarning: failed to clean up %s: %v (remove it manually)\n", runWorkDir, rmErr)
return
}
fmt.Printf("\ncleaned up %s; the run log is at %s\n", runWorkDir, filepath.Join(runStateDir, "state.json"))
}()
fmt.Println("--- preflight ---")
checker := preflight.New(preflight.Options{
BinaryPath: *binaryPath, ConfigPath: *configPath, DataDir: *dataDir, ContainerName: *containerName,
AdminURL: *adminURL, AdminUser: *adminUser, AdminPassword: *adminPassword,
TargetVersion: *targetVersion, MinFreeMultiple: *minFree, HTTPClient: httpClient,
CLIPath: *stalwartCLI, PythonPath: *pythonPath,
})
pfReport, err := checker.Run(ctx, store, rs)
fmt.Print(pfReport.String())
if err != nil {
return fmt.Errorf("preflight failed to complete: %w", err)
}
if pfReport.Blocking() {
return fmt.Errorf("preflight found blocking issues - see FAIL lines above")
}
p, err := plan.Decide(rs.SourceVersion, rs.TargetVersion)
if err != nil {
return fmt.Errorf("plan: %w", err)
}
fmt.Printf("\nplan: %s\n", p.Reason)
if !p.CrossesMajorBoundary {
fmt.Println("\nthis is a same-boundary patch upgrade: its settings don't need converting, " +
"so there is nothing to rehearse. A real run would be a binary swap and restart.")
return nil
}
scriptDest := filepath.Join(runWorkDir, "migrate_v016.py")
settingsPath := filepath.Join(runWorkDir, "settings.json")
principalsPath := filepath.Join(runWorkDir, "principals.json")
convertedConfig := filepath.Join(runWorkDir, "config.json")
convertedExport := filepath.Join(runWorkDir, "export.json")
unmigratedPath := filepath.Join(runWorkDir, "unmigrated.txt")
fmt.Println("\n--- dump (read-only) ---")
if _, err := store.RunStep(rs, checkpoint.PhaseBackup, "settings-dump", func() (checkpoint.StepOutcome, error) {
if err := os.MkdirAll(runWorkDir, 0o750); err != nil {
return checkpoint.StepOutcome{}, err
}
sum, err := backup.DownloadFile(ctx, httpClient, backup.DefaultMigrationScriptURL, scriptDest, *migrationScriptSHA256)
if err != nil {
return checkpoint.StepOutcome{}, err
}
pinNote := ""
if *migrationScriptSHA256 == "" {
pinNote = fmt.Sprintf(" (no pin configured - record sha256 %s as --migration-script-sha256 to pin it)", sum)
}
if err := backup.RunSettingsDump(ctx, backup.SettingsDumpOptions{
PythonPath: *pythonPath, ScriptPath: scriptDest, URL: *adminURL,
Username: *adminUser, Password: *adminPassword,
SettingsPath: settingsPath, PrincipalsPath: principalsPath,
}); err != nil {
return checkpoint.StepOutcome{}, err
}
_, settingsSize, err := backup.HashFile(settingsPath)
if err != nil {
return checkpoint.StepOutcome{}, err
}
_, principalsSize, err := backup.HashFile(principalsPath)
if err != nil {
return checkpoint.StepOutcome{}, err
}
return checkpoint.StepOutcome{
Detail: fmt.Sprintf("dumped settings (%d bytes) and principals (%d bytes) from %s%s",
settingsSize, principalsSize, *adminURL, pinNote),
}, nil
}); err != nil {
return fmt.Errorf("settings dump: %w", err)
}
fmt.Println(rs.Outcome(checkpoint.PhaseBackup, "settings-dump").Detail)
fmt.Println("\n--- convert ---")
if _, err := store.RunStep(rs, checkpoint.PhaseStage, "convert-settings", func() (checkpoint.StepOutcome, error) {
if err := backup.RunSettingsConvert(ctx, backup.SettingsConvertOptions{
PythonPath: *pythonPath, ScriptPath: scriptDest,
SettingsPath: settingsPath, PrincipalsPath: principalsPath,
ConfigPath: convertedConfig, OutputPath: convertedExport,
// migrate_v016.py writes unmigrated.txt into its working
// directory; without this it lands wherever the operator
// happened to be, or fails the convert if that isn't writable.
WorkDir: runWorkDir,
}); err != nil {
return checkpoint.StepOutcome{}, err
}
detail := "converted this instance's settings into a v0.16 apply plan"
if report, readErr := backup.ReadUnmigratedReport(unmigratedPath); readErr == nil && report != nil {
detail += fmt.Sprintf("; %d setting(s) will NOT carry over", report.TotalKeys)
}
return checkpoint.StepOutcome{Detail: detail}, nil
}); err != nil {
return fmt.Errorf("convert settings: %w", err)
}
if err := copyFile(convertedExport, keptPlan); err != nil {
fmt.Fprintf(os.Stderr, "warning: couldn't preserve the apply plan: %v\n", err)
} else if sum, size, hashErr := backup.HashFile(keptPlan); hashErr == nil {
rs.RecordArtifact("converted-export", checkpoint.Artifact{Path: keptPlan, SHA256: sum, SizeBytes: size})
fmt.Printf("apply plan: %s (%d bytes) - what WILL carry over\n", keptPlan, size)
}
unmigrated, err := backup.ReadUnmigratedReport(unmigratedPath)
if err != nil {
fmt.Fprintf(os.Stderr, "warning: couldn't read the unmigrated-settings report: %v\n", err)
} else if unmigrated != nil && unmigrated.TotalKeys > 0 {
if err := copyFile(unmigratedPath, keptWorklist); err != nil {
fmt.Fprintf(os.Stderr, "warning: couldn't preserve the worklist: %v\n", err)
} else {
unmigrated.Path = keptWorklist
if sum, size, hashErr := backup.HashFile(keptWorklist); hashErr == nil {
rs.RecordArtifact("unmigrated-settings", checkpoint.Artifact{Path: keptWorklist, SHA256: sum, SizeBytes: size})
}
}
// Categorised rather than counted: the raw total is alarming and
// misleading. Most of it either rebuilds itself or ships with
// v0.16 - see backup.Disposition.
fmt.Printf("\n !! %s\n", unmigrated.Classify().Summary(keptWorklist))
}
// Generate what we can of the gap (ARCHITECTURE.md §4.3). This is
// best-effort by design and reports its own coverage: the plan is
// worth having precisely because it is honest about how much of the
// worklist it does not touch.
fmt.Println("\n--- supplemental plan (best-effort) ---")
if err := generateSupplement(store, rs, settingsPath, principalsPath, unmigratedPath, keptSupplement); err != nil {
fmt.Fprintf(os.Stderr, "warning: couldn't generate the supplemental plan: %v\n", err)
}
if err := store.Save(rs); err != nil {
return fmt.Errorf("save run state: %w", err)
}
fmt.Printf("\nREHEARSAL COMPLETE for run %s. Nothing was modified: no data was copied, no server was started,\n"+
"and the store was never written to.\n", rs.RunID)
return nil
}
// copyFile duplicates src to dst, used to lift the worklist out of the
// scratch directory before it's cleaned up.
func copyFile(src, dst string) error {
if err := os.MkdirAll(filepath.Dir(dst), 0o750); err != nil {
return err
}
in, err := os.Open(src)
if err != nil {
return err
}
defer in.Close()
out, err := os.OpenFile(dst, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0o640)
if err != nil {
return err
}
defer out.Close()
if _, err := io.Copy(out, in); err != nil {
return err
}
return out.Sync()
}
// generateSupplement builds the best-effort apply plan for settings
// migrate_v016.py left behind, writes it beside the run's other
// conclusions, and reports honestly how much of the gap it closed.
//
// It is deliberately applied *after* export.json rather than merged into
// it: the official conversion is the authority on everything it handles,
// and a generated plan that overlapped it could silently override a
// correct mapping with a guessed one.
func generateSupplement(store *checkpoint.Store, rs *checkpoint.RunState, settingsPath, principalsPath, unmigratedPath, outPath string) error {
settings, err := backup.ReadSettingsDump(settingsPath)
if err != nil {
return err
}
unmigrated, err := backup.ReadUnmigratedKeys(unmigratedPath, settings)
if err != nil {
return err
}
plan, coverage, err := applyplan.Build(settings, unmigrated, applyplan.DefaultGenerators())
if err != nil {
return err
}
// Account roles don't live in the settings dump, so they come from the
// principals dump rather than through a settings Generator. Without
// this the migrated instance has no administrator: migrate_v016.py
// gives every account the User role, whatever it had before.
principals, err := backup.ReadPrincipalsDump(principalsPath)
if err != nil {
return err
}
roleOps, _, roleWarnings, err := applyplan.AccountRoleOperations(principals)
if err != nil {
return err
}
plan.Operations = append(plan.Operations, roleOps...)
coverage.Warnings = append(coverage.Warnings, roleWarnings...)
if len(roleOps) > 0 {
coverage.ObjectsByType["Account role"] += len(roleOps)
}
if len(plan.Operations) == 0 {
fmt.Println("nothing this tool can rebuild automatically yet - the whole worklist is manual")
return nil
}
if err := plan.WriteNDJSON(outPath); err != nil {
return err
}
if sum, size, hashErr := backup.HashFile(outPath); hashErr == nil {
rs.RecordArtifact("supplemental-plan", checkpoint.Artifact{Path: outPath, SHA256: sum, SizeBytes: size})
}
fmt.Printf("%s\n", coverage.Summary(6))
for _, w := range coverage.Warnings {
fmt.Printf(" warning: %s\n", w)
}
fmt.Printf("\n supplement: %s\n", outPath)
fmt.Println(" Review it, then apply it after export.json:")
fmt.Printf(" stalwart-cli apply --file %s --url <migrated-instance>\n", outPath)
fmt.Println(" It is generated, not authoritative - read it before you run it.")
_, err = store.RunStep(rs, checkpoint.PhaseStage, "supplemental-plan", func() (checkpoint.StepOutcome, error) {
return checkpoint.StepOutcome{
Detail: fmt.Sprintf("generated %d operation(s) covering %d of %d unmigrated setting(s)",
len(plan.Operations), coverage.CoveredKeys, coverage.TotalKeys),
}, nil
})
return err
}