Add snapshot retention with automatic pruning

Implements hybrid retention policy:
- Always keep 20 most recent snapshots
- Delete snapshots beyond #20 only if older than 180 days
- Genesis snapshot is always protected

Features:
- zfs-snap-prune script with --dry-run, --test, --verbose modes
- Comprehensive test suite (22 tests)
- Runs automatically after pacman operations
- Daily systemd timer for cleanup
- Regenerates GRUB menu after pruning

This prevents unbounded snapshot growth while preserving
recent history and the genesis snapshot.

1 files changed, 208 insertions, 0 deletions

diff --git a/custom/zfs-snap-prune b/custom/zfs-snap-prune
new file mode 100755
index 0000000..762ff99
--- /dev/null
+++ b/custom/zfs-snap-prune
@@ -0,0 +1,208 @@
+#!/bin/bash
+# zfs-snap-prune - Prune old ZFS snapshots with hybrid retention policy
+#
+# Retention Policy:
+#   - Always keep the N most recent snapshots (default: 20)
+#   - Delete snapshots beyond N only if older than MAX_AGE (default: 180 days)
+#   - Never delete genesis snapshot
+#
+# Usage:
+#   zfs-snap-prune [OPTIONS]
+#
+# Options:
+#   --dry-run       Show what would be deleted without deleting
+#   --verbose       Show decision for every snapshot
+#   --quiet         Suppress non-error output
+#   --test          Use mock data from stdin instead of real ZFS
+#   --help          Show this help message
+#
+# Environment variables:
+#   POOL_NAME       - ZFS pool name (default: zroot)
+#   ROOT_DATASET    - Root dataset path (default: ROOT/default)
+#   KEEP_COUNT      - Number of recent snapshots to always keep (default: 20)
+#   MAX_AGE_DAYS    - Delete older snapshots beyond KEEP_COUNT (default: 180)
+#   NOW_OVERRIDE    - Override current timestamp for testing (epoch seconds)
+
+set -e
+
+# Configuration (can be overridden by environment)
+POOL_NAME="${POOL_NAME:-zroot}"
+ROOT_DATASET="${ROOT_DATASET:-ROOT/default}"
+KEEP_COUNT="${KEEP_COUNT:-20}"
+MAX_AGE_DAYS="${MAX_AGE_DAYS:-180}"
+
+FULL_DATASET="${POOL_NAME}/${ROOT_DATASET}"
+
+# Flags
+DRY_RUN=false
+VERBOSE=false
+QUIET=false
+TEST_MODE=false
+
+# Colors for output
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+BLUE='\033[0;34m'
+NC='\033[0m'
+
+usage() {
+    sed -n '2,/^$/p' "$0" | sed 's/^# \?//'
+    exit 0
+}
+
+info() {
+    [[ "$QUIET" == "true" ]] && return
+    echo -e "${GREEN}[INFO]${NC} $1"
+}
+
+verbose() {
+    [[ "$VERBOSE" != "true" ]] && return
+    echo -e "${BLUE}[VERBOSE]${NC} $1"
+}
+
+warn() {
+    [[ "$QUIET" == "true" ]] && return
+    echo -e "${YELLOW}[WARN]${NC} $1"
+}
+
+error() {
+    echo -e "${RED}[ERROR]${NC} $1" >&2
+    exit 1
+}
+
+# Parse arguments
+while [[ $# -gt 0 ]]; do
+    case "$1" in
+        --dry-run)
+            DRY_RUN=true
+            shift
+            ;;
+        --verbose)
+            VERBOSE=true
+            shift
+            ;;
+        --quiet)
+            QUIET=true
+            shift
+            ;;
+        --test)
+            TEST_MODE=true
+            shift
+            ;;
+        --help|-h)
+            usage
+            ;;
+        *)
+            error "Unknown option: $1"
+            ;;
+    esac
+done
+
+# Check if running as root (skip in test mode)
+if [[ "$TEST_MODE" != "true" ]] && [[ $EUID -ne 0 ]]; then
+    error "This script must be run as root"
+fi
+
+# Get current timestamp (can be overridden for testing)
+NOW="${NOW_OVERRIDE:-$(date +%s)}"
+MAX_AGE_SECONDS=$((MAX_AGE_DAYS * 24 * 60 * 60))
+CUTOFF_TIME=$((NOW - MAX_AGE_SECONDS))
+
+info "Pruning snapshots for ${FULL_DATASET}"
+info "Policy: Keep ${KEEP_COUNT} recent, delete if older than ${MAX_AGE_DAYS} days"
+[[ "$DRY_RUN" == "true" ]] && info "DRY RUN - no changes will be made"
+
+# Get snapshots - either from ZFS or stdin (test mode)
+# Expected format: snapshot_name<TAB>creation_date_string
+# Example: zroot/ROOT/default@pre-pacman_2025-01-15	Wed Jan 15 10:30 2025
+if [[ "$TEST_MODE" == "true" ]]; then
+    # Read mock data from stdin
+    SNAPSHOTS=$(cat | tac)
+else
+    # Query real ZFS - sorted by creation (oldest first), then reversed for newest first
+    SNAPSHOTS=$(zfs list -H -t snapshot -o name,creation -s creation -r "$FULL_DATASET" 2>/dev/null | \
+        grep "^${FULL_DATASET}@" | \
+        tac) || true
+fi
+
+if [[ -z "$SNAPSHOTS" ]]; then
+    info "No snapshots found"
+    exit 0
+fi
+
+# Count snapshots
+TOTAL=$(echo "$SNAPSHOTS" | wc -l)
+info "Found ${TOTAL} snapshots"
+
+# Track results
+DELETED=0
+KEPT=0
+POSITION=0
+
+# Process each snapshot
+while IFS=$'\t' read -r snapshot creation_str; do
+    [[ -z "$snapshot" ]] && continue
+
+    POSITION=$((POSITION + 1))
+    SNAP_NAME="${snapshot##*@}"
+
+    # Parse creation time
+    if [[ "$TEST_MODE" == "true" ]]; then
+        # In test mode, creation_str is epoch seconds
+        SNAP_TIME="$creation_str"
+    else
+        # In real mode, parse date string
+        SNAP_TIME=$(date -d "$creation_str" +%s 2>/dev/null || echo "0")
+    fi
+
+    AGE_DAYS=$(( (NOW - SNAP_TIME) / 86400 ))
+
+    # Decision logic
+    if [[ $POSITION -le $KEEP_COUNT ]]; then
+        # Always keep the first KEEP_COUNT snapshots (most recent)
+        verbose "KEEP: ${SNAP_NAME} (position ${POSITION}/${KEEP_COUNT}, ${AGE_DAYS} days old) - within keep count"
+        KEPT=$((KEPT + 1))
+    elif [[ "$SNAP_NAME" == "genesis" ]]; then
+        # Never delete genesis
+        verbose "KEEP: ${SNAP_NAME} (position ${POSITION}, ${AGE_DAYS} days old) - genesis protected"
+        KEPT=$((KEPT + 1))
+    elif [[ $SNAP_TIME -ge $CUTOFF_TIME ]]; then
+        # Not old enough to delete
+        verbose "KEEP: ${SNAP_NAME} (position ${POSITION}, ${AGE_DAYS} days old) - younger than ${MAX_AGE_DAYS} days"
+        KEPT=$((KEPT + 1))
+    else
+        # Delete: beyond keep count AND older than max age
+        if [[ "$DRY_RUN" == "true" ]]; then
+            info "WOULD DELETE: ${SNAP_NAME} (position ${POSITION}, ${AGE_DAYS} days old)"
+            DELETED=$((DELETED + 1))
+        elif [[ "$TEST_MODE" == "true" ]]; then
+            # Test mode: simulate deletion (don't actually call zfs)
+            verbose "DELETE: ${SNAP_NAME} (position ${POSITION}, ${AGE_DAYS} days old)"
+            DELETED=$((DELETED + 1))
+        else
+            verbose "DELETE: ${SNAP_NAME} (position ${POSITION}, ${AGE_DAYS} days old)"
+            if zfs destroy "$snapshot" 2>/dev/null; then
+                DELETED=$((DELETED + 1))
+            else
+                warn "Failed to delete ${snapshot}"
+            fi
+        fi
+    fi
+done <<< "$SNAPSHOTS"
+
+# Summary
+info "Summary: ${KEPT} kept, ${DELETED} deleted"
+
+# Regenerate GRUB menu if we deleted anything (skip in dry-run and test modes)
+if [[ $DELETED -gt 0 ]] && [[ "$DRY_RUN" != "true" ]] && [[ "$TEST_MODE" != "true" ]]; then
+    if [[ -x /usr/local/bin/grub-zfs-snap ]]; then
+        info "Regenerating GRUB menu..."
+        /usr/local/bin/grub-zfs-snap
+    fi
+fi
+
+# Exit with special code for testing (number of deleted)
+if [[ "$TEST_MODE" == "true" ]]; then
+    echo "RESULT:kept=${KEPT},deleted=${DELETED}"
+fi