Files
selfpost/internal/health/processes.go
T
mixandClaude Opus 5.5 c171fb5bc7
test / test (push) Canceled after 0s
from-header-binding: the sender check, what it costs, its pages; guide, architecture, security
A message's From: line is held against the application that sends it, under
the rule Postfix holds the envelope sender to. OpenDKIM chooses the key by
that line, so until now an application could have mail signed as another
domain of the server or as the server itself.

The check (internal/milter/sender.go)
- A role of the panel on a socket of its own (SENDER_CHECK_SOCKET), first in
  smtpd_milters and strict: default_action=tempfail. The journal milter keeps
  its socket and stays fail-open. postfix-wrapper waits for the new socket.
- It reads no database: what an application may send as comes from Postfix's
  own sender_login_maps, what the server signs for from OpenDKIM's
  SigningTable. Both are read again when they change; one that cannot be
  read defers mail.
- One From: field, one address, allowed by the envelope's lookup; then no
  other name of the SigningTable may appear anywhere in the raw line.
  Refused: 550 5.7.1 naming the line, reduced to printable ASCII.
- A refusal is a send-log row with reason "from" and the header address
  (migration 0006), written by another goroutine and not waited for. The
  journal's refusals record their reason too (limit, client), and the
  notifier takes only a limit's.

What it costs (docs/plans/from-header-binding.md, Statistics)
- The role marks itself alive every 2 s; at start the gap since the mark is
  downtime, counted from Postfix's start if later and not at all when
  supervisord says Postfix is not running.
- The log tailer counts the connections Postfix turned away: a warning that
  names the socket, then milter-reject "Service unavailable" from the same
  smtpd. Seen on the stand: the client is refused from CONNECT on and leaves
  at AUTH (454), so no MAIL line is ever written.
- sender_check_days, a row per UTC day, kept a year.

Pages
- Health: the card "Sender check", the socket's row, and the cards layout -
  boxes numbered by importance in two stacks read in rows; the left stack
  polls, the three readings of the right one are swapped in by id.
- Outbound log: a refused row says why in its tag (wrong From, wrong client,
  rate-limited); the message's page shows the From: line and what the
  application may send as.
- Help: topic "Sender check", word for word the mockup's.

Found by the stand, fixed here: the names for the second look came from the
sender bindings, and a domain with a key and no application yet was not
among them.

e2e: negative_journal_milter_fail_open is gone - it proved that mail passes
with the panel down, which is no longer so. In its place
sender_check_holds_the_from_line (another domain's From:, the hostname's, a
stranger's, an address off the list, another sender in the display name -
refused; own addresses - delivered) and panel_down_defers_mail (4xx with the
panel stopped, the container alive, mail again after, downtime and one
deferred message on Health, the card warning).

Test server, this tree before the last cosmetic edits (a fixture row, a
fact's size, docs): go vet ok, go test 23 packages ok, make e2e ok (337 s).
Locally after them: internal/web/view ok.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-12 19:35:43 +03:00

141 lines
4.3 KiB
Go

package health
import (
"fmt"
"os/exec"
"strconv"
"strings"
"time"
)
// supervisorConf is the supervisord configuration the panel's other control
// calls already address (see internal/postfix and internal/domain).
const supervisorConf = "/etc/supervisor/supervisord.conf"
// oneShotPrograms are supervisord entries that are meant to sit idle: they are
// started on demand and exit immediately, so STOPPED/EXITED is their healthy
// state rather than a fault (see build/supervisord.conf).
var oneShotPrograms = map[string]bool{
"postfix-reload": true,
}
// Process is one supervised program as supervisord reports it.
type Process struct {
Name string
State string // supervisord's own state word, e.g. RUNNING
Detail string // the rest of the line: pid/uptime, or exit information
Status Status
// OneShot marks a program that is started on demand and exits at once
// (oneShotPrograms): idle is its healthy state, and it is not one of the
// programs that are expected to be running.
OneShot bool
}
// Processes returns the state of every supervised program (architecture.md §
// Image and processes — the three processes plus the reload/cert/logrotate
// helpers).
//
// The command takes fixed arguments and no user input, so it never goes through
// a shell (security.md). `supervisorctl status` deliberately exits non-zero when
// some program is not running, so the output is parsed first and the exit status
// only matters when nothing could be parsed from it.
func Processes() ([]Process, error) {
cmd := exec.Command("supervisorctl", "-c", supervisorConf, "status")
out, err := cmd.CombinedOutput()
procs := parseProcesses(string(out))
if len(procs) == 0 {
if err != nil {
return nil, fmt.Errorf("supervisorctl status: %w: %s", err, strings.TrimSpace(string(out)))
}
return nil, fmt.Errorf("supervisorctl status: no programs reported")
}
return procs, nil
}
// supervisorStates are the state words supervisord prints. Lines whose second
// field is not one of them are not status lines (banners, error text) and are
// skipped, so unexpected output cannot masquerade as a process.
var supervisorStates = map[string]bool{
"STOPPED": true,
"STARTING": true,
"RUNNING": true,
"BACKOFF": true,
"STOPPING": true,
"EXITED": true,
"FATAL": true,
"UNKNOWN": true,
}
// parseProcesses turns supervisorctl's tabular output into Process values. Each
// status line is "<name> <STATE> <detail...>", column-aligned with spaces.
func parseProcesses(out string) []Process {
var procs []Process
for _, line := range strings.Split(out, "\n") {
fields := strings.Fields(line)
if len(fields) < 2 || !supervisorStates[fields[1]] {
continue
}
name, state := fields[0], fields[1]
procs = append(procs, Process{
Name: name,
State: state,
Detail: strings.Join(fields[2:], " "),
Status: processStatus(name, state),
OneShot: oneShotPrograms[name],
})
}
return procs
}
// Uptime reads for how long a running program has been up from the detail
// supervisord prints for it: "pid 41, uptime 0:03:12", or with days in front,
// "pid 41, uptime 18 days, 3:02:11".
func Uptime(detail string) (time.Duration, bool) {
_, rest, found := strings.Cut(detail, "uptime ")
if !found {
return 0, false
}
var days int
if d, clock, hasDays := strings.Cut(rest, ", "); hasDays {
n, _, _ := strings.Cut(d, " ")
v, err := strconv.Atoi(n)
if err != nil || v < 0 {
return 0, false
}
days, rest = v, clock
}
parts := strings.Split(strings.TrimSpace(rest), ":")
if len(parts) != 3 {
return 0, false
}
var hms [3]int
for i, p := range parts {
v, err := strconv.Atoi(p)
if err != nil || v < 0 {
return 0, false
}
hms[i] = v
}
return time.Duration(days)*24*time.Hour + time.Duration(hms[0])*time.Hour +
time.Duration(hms[1])*time.Minute + time.Duration(hms[2])*time.Second, true
}
// processStatus grades a supervisord state. A one-shot program that is not
// running is healthy; anything else that is not RUNNING means a component of
// the mail path is down or flapping.
func processStatus(name, state string) Status {
switch state {
case "RUNNING":
return StatusOK
case "STARTING", "STOPPING":
return StatusWarn
case "STOPPED", "EXITED":
if oneShotPrograms[name] {
return StatusOK
}
return StatusError
default: // BACKOFF, FATAL, UNKNOWN
return StatusError
}
}