From 244a73b7a2b38a17e9331606fab6551689553098 Mon Sep 17 00:00:00 2001
From: Mike Reeves <reevesmk@gmail.com>
Date: Fri, 15 May 2026 08:48:54 -0400
Subject: [PATCH] Make so-postgres-backup fail-safe against silent corruption

The dump pipeline returned gzip's exit status, so a pg_dumpall that
died mid-stream still produced a valid .gz holding a truncated dump,
written straight to the final filename. The idempotency check then
blocked retries for the day and the corrupt file counted toward
retention, evicting a good backup each day until none remained.

- set -o pipefail so a failed pg_dumpall fails the pipeline
- dump to a .tmp file and atomically rename only after success, so
  the final filename appears only for a complete backup
- gzip -t integrity check before publishing
- trap-based cleanup of the temp file; sweep stale temps at startup
- run retention only after a successful backup, with a glob
  restricted to finished backups
- log timestamped OK/ERROR outcomes to /opt/so/log/postgres/backup.log
---
 salt/postgres/tools/sbin/so-postgres-backup | 48 ++++++++++++++++++---
 1 file changed, 41 insertions(+), 7 deletions(-)

diff --git a/salt/postgres/tools/sbin/so-postgres-backup b/salt/postgres/tools/sbin/so-postgres-backup
index 9db522336..08a73e3a4 100644
--- a/salt/postgres/tools/sbin/so-postgres-backup
+++ b/salt/postgres/tools/sbin/so-postgres-backup
@@ -7,15 +7,29 @@
 
 . /usr/sbin/so-common
 
+# Without pipefail, a pipeline's exit status is gzip's. A failed pg_dumpall would
+# otherwise be masked by a successful gzip, silently producing a valid .gz that
+# holds a truncated dump.
+set -o pipefail
+
 # Backups contain role password hashes and full chat data; keep them 0600.
 umask 0077
 
 TODAY=$(date '+%Y_%m_%d')
 BACKUPDIR=/nsm/backup
 BACKUPFILE="$BACKUPDIR/so-postgres-backup-$TODAY.sql.gz"
+TMPFILE="$BACKUPFILE.tmp"
 MAXBACKUPS=7
+LOGFILE=/opt/so/log/postgres/backup.log
 
-mkdir -p $BACKUPDIR
+log() {
+  echo "$(date '+%Y-%m-%d %H:%M:%S') $*" >> "$LOGFILE"
+}
+
+mkdir -p "$BACKUPDIR"
+
+# Remove any temp files left behind by a previously crashed run
+rm -f "$BACKUPDIR"/so-postgres-backup-*.sql.gz.tmp
 
 # Skip if already backed up today
 if [ -f "$BACKUPFILE" ]; then
@@ -27,13 +41,33 @@ if ! docker ps --format '{{.Names}}' | grep -q '^so-postgres$'; then
   exit 0
 fi
 
-# Dump all databases and roles, compress
-docker exec so-postgres pg_dumpall -U postgres | gzip > "$BACKUPFILE"
+# Always clean up the temp file on exit; the success path clears this trap
+# after the atomic rename so the finished backup is not deleted.
+trap 'rm -f "$TMPFILE"' EXIT
 
-# Retention cleanup
-NUMBACKUPS=$(find $BACKUPDIR -type f -name "so-postgres-backup*" | wc -l)
+# Dump all databases and roles, compress. Write to a temp file so the final
+# filename only ever appears for a complete, verified backup.
+if ! docker exec so-postgres pg_dumpall -U postgres | gzip > "$TMPFILE"; then
+  log "ERROR: pg_dumpall/gzip failed; backup aborted"
+  exit 1
+fi
+
+# Verify the compressed stream is intact before publishing it
+if ! gzip -t "$TMPFILE"; then
+  log "ERROR: backup failed gzip integrity check; backup aborted"
+  exit 1
+fi
+
+# Atomically publish the verified backup
+mv "$TMPFILE" "$BACKUPFILE"
+trap - EXIT
+log "OK: wrote $BACKUPFILE"
+
+# Retention cleanup (only reached after a successful backup). The glob is
+# restricted to finished backups so an in-progress .tmp can never be counted.
+NUMBACKUPS=$(find "$BACKUPDIR" -type f -name "so-postgres-backup-*.sql.gz" | wc -l)
 while [ "$NUMBACKUPS" -gt "$MAXBACKUPS" ]; do
-  OLDEST=$(find $BACKUPDIR -type f -name "so-postgres-backup*" -printf '%T+ %p\n' | sort | head -n 1 | awk -F" " '{print $2}')
+  OLDEST=$(find "$BACKUPDIR" -type f -name "so-postgres-backup-*.sql.gz" -printf '%T+ %p\n' | sort | head -n 1 | awk -F" " '{print $2}')
   rm -f "$OLDEST"
-  NUMBACKUPS=$(find $BACKUPDIR -type f -name "so-postgres-backup*" | wc -l)
+  NUMBACKUPS=$(find "$BACKUPDIR" -type f -name "so-postgres-backup-*.sql.gz" | wc -l)
 done