Files
Felis/cmd/felis/update.go
flyemoji ecbeb20761 fix(bootstrap): reuse the installed root domain instead of re-deriving it
detect_node_ip recomputed FELIS_ROOT_DOMAIN from scratch on every run and fell
back to <node-ip>.nip.io. Nothing read the domain back out of the felis.toml an
earlier run wrote, so it survived only as long as the operator kept passing the
same environment.

That made re-running the installer destructive on any install with a real
domain, and re-running it is not optional: it is the only way to move felis-api
to a newer release, which is what `felis update` points operators at. A bare
re-run rewrote root_domain, panel_hostname and admin_hostname to nip.io names
while ensure_panel_tls_cert returned early on the certificate it had already
written, leaving the console serving a cert for hostnames it no longer answered
to -- with no re-domain flow to recover through.

Precedence is now explicit FELIS_ROOT_DOMAIN, then the domain the last run
persisted, then the nip.io default. First installs are unaffected. Deliberate
re-domains still work, because there is no other route to one, but they now warn
that the write-once certificate is not reissued and that the proxy and login
config carry the old name too.

Secrets were never exposed to this: load_or_make_secrets has always sourced
secrets.env before generating anything. The domain was the one piece of install
identity with no read-back.

The channel is deliberately left alone. FELIS_VERSION_BOOTSTRAP is not persisted
either, but defaulting a re-run to the release channel installs a working build
rather than breaking one, so cmd/felis/update.go states that instead. Its
warning about the domain went with the bug and would now be false.

Verified against the shipped function text: the ladder holds for a fresh host,
a re-run with and without the variable set, a re-domain, and a felis.toml whose
root_domain is missing or empty. Reverting the one line reproduces the nip.io
overwrite.
2026-07-20 21:20:08 +09:00

276 lines
12 KiB
Go

package main
import (
"context"
"flag"
"fmt"
"io"
"sort"
"strings"
"time"
"felis.lolicon.best/internal/updater"
"felis.lolicon.best/internal/updates"
)
// updateTimeout bounds the whole discovery pass. Each release source already caps
// its own HTTP client, but a hung DNS or a stalled TLS handshake would otherwise
// leave the operator staring at a silent terminal.
const updateTimeout = 60 * time.Second
// updateTarget maps a user-facing selector (`--panel`) onto the planner's component
// name and the command that actually performs the update.
//
// The apply side is deliberately NOT implemented in this command. Every component
// here is installed by deploy/bootstrap.sh, which is idempotent, already handles the
// parts that are easy to get wrong (Velocity's pinned MINOR, the atomic jar install,
// the k3s image re-import that a byte-identical StatefulSet template will not
// trigger on its own), and is the path that gets exercised on every install. A
// second installer living in this file would duplicate that policy, could drift from
// it silently, and would be reachable only on a live node where a mistake takes the
// proxy or the control plane down. So `felis update` reports, and hands the operator
// the tested command — it does not re-implement it.
type updateTarget struct {
// selector is the flag name without dashes.
selector string
// component is the updates planner's name for this piece, or "" when the planner
// deliberately does not track it (Minecraft, which is pinned).
component string
// note explains what this selector covers, printed above the command.
note string
// command is the exact, already-tested way to apply it.
command string
}
// updateTargets is the selector table. panel and plugins both resolve to felis-api
// because they are not separately versioned: the panel is compiled into the felis
// binary with //go:embed, and the plugin jars are built from this same repo in the
// same bootstrap run, so all three move together and carry one version.
var updateTargets = []updateTarget{
{
selector: "panel",
component: "felis-api",
note: "the panel is embedded in the felis binary (//go:embed), so updating it means rebuilding the felis image and rolling felis-api",
command: "sudo felis setup",
},
{
selector: "velocity",
component: "velocity",
note: "re-runs install_velocity: newest BUILD of the pinned minor (FELIS_VELOCITY_VERSION), atomic jar install, then restarts felis-velocity",
command: "sudo felis setup",
},
{
selector: "plugins",
component: "felis-api",
note: "felis-velocity.jar is a host-file swap, but felis-paper.jar and felis-limbo.jar are baked into the lobby/limbo images and need a rebuild + k3s image re-import",
command: "sudo felis setup",
},
{
selector: "mc",
component: "", // never tracked: see the pin note below
note: "Minecraft is pinned by policy (\"能不动的就别动\") and Felis never proposes a version change for it. A server's version is a property of that server's image — change it on the server, not through a platform update",
command: "",
},
}
// cmdUpdate reports what can be updated and what is already current.
//
// Bare `felis update` prints the status of every tracked component. Selector flags
// (--panel/--velocity/--mc/--plugins/--all) narrow that report to the components
// they name AND print how to apply each one. --force additionally prints the apply
// instruction for a selected component that is already up to date, for the
// reinstall/repair case.
//
// It never applies anything and never mutates the node, so unlike setup/breakGlass
// it needs no root. The versions it reads come from this host: k3s and cloudflared
// answer `--version`, Velocity's version is read out of the installed jar's
// manifest, and felis-api's is this binary's own build stamp — the same value
// `felis version` prints, which is what the user asked to be the source of truth.
func cmdUpdate(args []string, stdout, stderr io.Writer) int {
fs := flag.NewFlagSet("update", flag.ContinueOnError)
fs.SetOutput(stderr)
panel := fs.Bool("panel", false, "select the panel + control plane (felis-api)")
velocity := fs.Bool("velocity", false, "select the Velocity proxy")
mc := fs.Bool("mc", false, "select Minecraft (pinned; reported only)")
plugins := fs.Bool("plugins", false, "select the Felis plugin jars (velocity/paper/limbo)")
all := fs.Bool("all", false, "select every component above")
force := fs.Bool("force", false, "print the apply command for a selected component even when it is already up to date")
velocityJar := fs.String("velocity-jar", updater.DefaultVelocityJarPath, "path to the installed Velocity jar to read the current version from")
if err := fs.Parse(args); err != nil {
return 2
}
if fs.NArg() > 0 {
fmt.Fprintf(stderr, "felis update: unexpected argument %q (this command takes flags only)\n", fs.Arg(0))
return 2
}
selected := map[string]bool{}
for sel, on := range map[string]bool{"panel": *panel, "velocity": *velocity, "mc": *mc, "plugins": *plugins} {
if on || *all {
selected[sel] = true
}
}
ctx, cancel := context.WithTimeout(context.Background(), updateTimeout)
defer cancel()
rn := &updater.Runner{
Gatherer: updater.NewHostGatherer(resolvedVersion(), *velocityJar),
Source: updater.NewRoutingSource(updater.Topology()),
// Notifier and Applier stay nil on purpose: a human typing this command IS the
// notification, and nothing here applies. The zero Window below means every
// Scheduled component degrades to a notify, so the report can never claim an
// apply is under way.
}
res, err := rn.Run(ctx, time.Now(), updates.Window{})
if err != nil {
fmt.Fprintf(stderr, "felis update: %v\n", err)
return 1
}
fmt.Fprint(stdout, renderUpdateReport(res, selected))
if len(selected) > 0 {
fmt.Fprint(stdout, renderApplyGuidance(res, selected, *force))
}
return 0
}
// renderUpdateReport renders the component status table. With no selectors it shows
// every tracked component; with selectors it shows only the components those
// selectors name, so `felis update --velocity` is a focused answer rather than the
// whole platform. Components whose current version could not be read are listed
// separately rather than silently dropped — a component missing from the table with
// no explanation reads as "fine", which is the one thing it is not.
func renderUpdateReport(res updater.Result, selected map[string]bool) string {
plan := res.RunResult.Plan
if len(selected) > 0 {
want := map[string]bool{}
for _, t := range updateTargets {
if selected[t.selector] && t.component != "" {
want[t.component] = true
}
}
filtered := make([]updates.Action, 0, len(plan))
for _, a := range plan {
if want[a.Component] {
filtered = append(filtered, a)
}
}
plan = filtered
}
var b strings.Builder
if len(plan) > 0 {
b.WriteString(updates.Report(plan))
}
// Two different failures both end up as a component the report cannot speak to,
// and both are printed rather than swallowed. A component that simply vanishes
// from the table reads as "fine", and a bare "latest unknown" line reads as "there
// is nothing newer" — when the truth may be that the release feed was unreachable.
// Naming the cause is the difference between a report and a guess.
writeErrs(&b, "current version unreadable", res.GatherErrors, selected)
writeErrs(&b, "latest version undiscoverable", res.RunResult.SourceErrors, selected)
// With selectors active an empty table is not a surprise — `--mc` names a component
// the planner deliberately does not track — and the guidance below already explains
// it, so stay quiet rather than printing a bare "nothing matched" that reads as an
// error. Without selectors an empty table means nothing is tracked at all, which
// does need saying.
if b.Len() == 0 && len(selected) == 0 {
return "No components tracked.\n"
}
return b.String()
}
// writeErrs appends one explanatory line per failed component, in a stable order so
// the output does not shuffle between runs, honouring the active selector filter.
func writeErrs(b *strings.Builder, label string, errs map[string]error, selected map[string]bool) {
names := make([]string, 0, len(errs))
for name := range errs {
names = append(names, name)
}
sort.Strings(names)
for _, name := range names {
if len(selected) > 0 && !selectedCovers(selected, name) {
continue
}
fmt.Fprintf(b, "%-13s %s: %v\n", name, label, errs[name])
}
}
// selectedCovers reports whether any chosen selector maps to this component.
func selectedCovers(selected map[string]bool, component string) bool {
for _, t := range updateTargets {
if selected[t.selector] && t.component == component {
return true
}
}
return false
}
// renderApplyGuidance prints, for each selected target, how to actually apply the
// update. A target that is already current is skipped unless --force was passed.
func renderApplyGuidance(res updater.Result, selected map[string]bool, force bool) string {
byComponent := map[string]updates.Action{}
for _, a := range res.RunResult.Plan {
byComponent[a.Component] = a
}
var b strings.Builder
var offeredCommand bool
var offeredFelisAPI bool
for _, t := range updateTargets {
if !selected[t.selector] {
continue
}
// Minecraft has no planner entry by design; state the pin and move on. There is
// no command to offer, so this must not arm the trailer below.
if t.component == "" {
fmt.Fprintf(&b, "\n--%s: %s.\n", t.selector, t.note)
continue
}
a, planned := byComponent[t.component]
// "Up to date" requires actually KNOWING the latest version. ActionNone covers
// both "nothing newer exists" and "the release feed could not be read", and
// collapsing those would report an unreachable upstream as currency — telling an
// operator they are current when nobody checked is the one answer an update tool
// must never give. LatestKnown is what separates them.
if planned && a.Kind == updates.ActionNone && a.LatestKnown && !force {
fmt.Fprintf(&b, "\n--%s: %s is already up to date; nothing to apply (use --force to reinstall anyway).\n", t.selector, t.component)
continue
}
fmt.Fprintf(&b, "\n--%s: %s\n", t.selector, t.note)
if planned && !a.LatestKnown {
// Unknown latest still offers the command: the operator asked about this
// component, and reinstalling the current release is a valid repair action.
fmt.Fprintf(&b, " note: cannot tell whether %s is current — its latest version could not be discovered (see above); this reinstalls it either way\n", t.component)
}
fmt.Fprintf(&b, " run: %s\n", t.command)
offeredCommand = true
offeredFelisAPI = offeredFelisAPI || t.component == "felis-api"
}
// Only explain the command when one was actually offered; a --mc-only run has
// nothing to run and the trailer would be a non-sequitur.
if offeredCommand {
b.WriteString("\nfelis setup is idempotent and re-runs the installer that owns these components;\nit does not reinstall what is already current. Restart game servers afterwards.\n")
}
// Scoped to felis-api because it is the only component setup cannot move forward.
// velocity is fine: install_velocity re-resolves the newest build of the pinned minor
// on every run. But setup hands deploy/bootstrap.sh the binary it is itself running
// (FELIS_BOOTSTRAP_BINARY), and that arm skips the release lookup entirely, so it
// rebuilds the image and rolls the deployment from the SAME binary -- a run that looks
// like a successful update and leaves the version unchanged.
//
// The installer is the only thing that moves felis-api. It is safe to point at now
// that detect_node_ip reuses the installed root domain, so what is left to warn about
// is the channel: FELIS_VERSION_BOOTSTRAP is not persisted anywhere and defaults to
// release, so a bare re-run on a host tracking main quietly moves it onto releases.
// That is a channel change, not a broken install, which is why it is one clause and
// not a paragraph.
if offeredFelisAPI {
b.WriteString("\nfelis-api (panel, plugins) is the exception: setup re-images it from the felis binary\nalready on this host, so it cannot install a NEWER felis-api. Re-run the bootstrap\ninstaller for that -- it keeps this install's root domain. It does default to the\nrelease channel, so pass FELIS_VERSION_BOOTSTRAP=dev if this host tracks main.\n")
}
return b.String()
}