#!/usr/bin/env bash
# Monitor CIFS mounts from /etc/fstab and (re)mount if needed.
# Designed for cron/systemd timer. Requires root.

set -Eeuo pipefail
IFS=$'\n\t'

# ---------------------------
# Config (edit if desired)
# ---------------------------
LOGFILE="/var/log/cifs-remount.log"     # Set empty ("") to disable file logging.
PING_COUNT=1
PING_TIMEOUT=1                # seconds
TCP_TIMEOUT=2                 # seconds for port 445 check
PROBE_TIMEOUT=2             # seconds to test mount health (listing)
REMOUNT_RETRY=2             # attempts
SLEEP_BETWEEN=1             # seconds between retries
SYSLOG_TAG="cifs-watch"

DRY_RUN=0
VERBOSE=0

# ---------------------------
# Helpers
# ---------------------------
log() {
    local level="$1"; shift
    local msg="$*"
    local ts
    ts="$(date '+%Y-%m-%d %H:%M:%S')"
    if [[ "$VERBOSE" -eq 1 || "$level" != "DEBUG" ]]; then
        echo "[$ts] [$level] $msg"
    fi
    logger -t "$SYSLOG_TAG[$$]" -p "user.$(tr '[:upper:]' '[:lower:]' <<<"$level")" -- "$msg" || true
    if [[ -n "$LOGFILE" ]]; then
        ( umask 0077; echo "[$ts] [$level] $msg" >> "$LOGFILE" ) || true
    fi
}

vdbg() { [[ "$VERBOSE" -eq 1 ]] && log "DEBUG" "$*"; }
fail() { log "ERROR" "$*"; exit 1; }
have_cmd() { command -v "$1" >/dev/null 2>&1; }

tcp_open_445() {
    local host="$1"
    if have_cmd nc; then
        nc -z -w "$TCP_TIMEOUT" "$host" 445 >/dev/null 2>&1
        return $?
    else
        timeout "$TCP_TIMEOUT" bash -c "cat < /dev/null > /dev/tcp/$host/445" 2>/dev/null
        return $?
    fi
}

host_reachable() {
    local host="$1"
    if ! getent ahosts "$host" >/dev/null 2>&1; then
        vdbg "DNS resolution failed for $host"
        return 1
    fi
    if have_cmd ping; then
        ping -c "$PING_COUNT" -W "$PING_TIMEOUT" "$host" >/dev/null 2>&1 || vdbg "Ping to $host failed"
    fi
    if tcp_open_445 "$host"; then
        return 0
    else
        vdbg "TCP/445 closed on $host"
        return 1
    fi
}

is_cifs_mounted() {
    local mnt="$1"
    if findmnt -no FSTYPE -T "$mnt" 2>/dev/null | grep -qi '^cifs$'; then
        return 0
    fi
    return 1
}

mount_healthy() {
    local mnt="$1"
    timeout "$PROBE_TIMEOUT" bash -c 'ls -1A -- "$0" >/dev/null 2>&1' "$mnt"
}

repair_mount() {
    local mnt="$1"
    local attempt=1
    while (( attempt <= REMOUNT_RETRY )); do
        vdbg "Attempt $attempt: remounting $mnt"
        if (( DRY_RUN )); then
            log "INFO" "DRY-RUN: would remount $mnt"
            return 0
        fi
        if mount -o remount "$mnt" >/dev/null 2>&1; then
            if mount_healthy "$mnt"; then
                log "INFO" "Remounted healthy: $mnt"
                return 0
            fi
            vdbg "Remount completed but health probe failed: $mnt"
        fi
        sleep "$SLEEP_BETWEEN"
        (( attempt++ ))
    done

    log "WARN" "Remount failed/unhealthy for $mnt; trying forced unmount + clean mount"
    if (( DRY_RUN )); then
        log "INFO" "DRY-RUN: would umount -f $mnt && mount $mnt"
        return 0
    fi

    if umount -f "$mnt" >/dev/null 2>&1 || umount -l "$mnt" >/dev/null 2>&1; then
        :
    else
        log "WARN" "Unable to unmount $mnt; will still attempt a mount"
    fi

    if mount "$mnt" >/dev/null 2>&1; then
        if mount_healthy "$mnt"; then
            log "INFO" "Mounted healthy: $mnt"
            return 0
        else
            log "WARN" "Mounted but health probe failed: $mnt"
            return 1
        fi
    else
        log "ERROR" "Mount failed for $mnt"
        return 1
    fi
}

usage() {
    cat <<'USAGE'
cifs-watch.sh [-n|--dry-run] [-v|--verbose] [--logfile PATH]

Monitors CIFS entries in /etc/fstab, checks server reachability, and (re)mounts as needed.
- Processes uncommented lines with type "cifs".
- Accepts fstab lines with 4–6 fields.
- Skips entries containing "noauto".

Options:
    -n, --dry-run         Show actions without changing anything
    -v, --verbose         More detailed output
            --logfile P     Override logfile path (empty to disable file logging)
USAGE
}

# ---------------------------
# Parse args
# ---------------------------
while [[ $# -gt 0 ]]; do
    case "$1" in
        -n|--dry-run) DRY_RUN=1; shift ;;
        -v|--verbose) VERBOSE=1; shift ;;
        --logfile) LOGFILE="${2:-}"; shift 2 ;;
        -h|--help) usage; exit 0 ;;
        *) echo "Unknown arg: $1"; usage; exit 2 ;;
    esac
done

# Ensure tools present
for bin in findmnt mount umount awk grep sed timeout; do
    have_cmd "$bin" || fail "Required command not found: $bin"
done

# ---------------------------
# Main: parse /etc/fstab
# ---------------------------
mapfile -t CIFS_LINES < <(awk '
    $0 !~ /^[[:space:]]*#/ && NF>=4 && tolower($3)=="cifs" { print }
' /etc/fstab)

if [[ ${#CIFS_LINES[@]} -eq 0 ]]; then
    log "INFO" "No CIFS entries found in /etc/fstab. Nothing to do."
    exit 0
fi

overall_rc=0

for line in "${CIFS_LINES[@]}"; do
    # fields: fs_spec    mountpoint    fstype    options    [dump]    [pass]
    fs_spec=$(awk '{print $1}' <<<"$line")
    mnt_point=$(awk '{print $2}' <<<"$line")
    fstype=$(awk '{print tolower($3)}' <<<"$line")
    options=$(awk '{print $4}' <<<"$line")
    dumpv=$(awk 'NF>=5{print $5}' <<<"$line")
    passv=$(awk 'NF>=6{print $6}' <<<"$line")

    # Skip noauto entries
    if grep -qi '(^|,)noauto(,|$)' <<<",$options,"; then
        vdbg "Skipping noauto CIFS entry: $mnt_point ($fs_spec)"
        continue
    fi

    # Parse server from //server/share
    if [[ "$fs_spec" =~ ^//([^/]+)/.+$ ]]; then
        server="${BASH_REMATCH[1]}"
    else
        log "WARN" "Could not parse server from fs_spec: $fs_spec (skipping)"
        continue
    fi

    log "INFO" "Checking CIFS mount: $mnt_point (server: $server)"

    if is_cifs_mounted "$mnt_point"; then
        if mount_healthy "$mnt_point"; then
            vdbg "Healthy: $mnt_point"
            continue
        else
            log "WARN" "Mounted but unhealthy: $mnt_point"
        fi
    else
        log "WARN" "Not mounted: $mnt_point"
    fi

    if host_reachable "$server"; then
        log "INFO" "Server reachable: $server — attempting repair for $mnt_point"
    else
        log "ERROR" "Server NOT reachable: $server — skipping $mnt_point for now"
        overall_rc=1
        continue
    fi

    if ! repair_mount "$mnt_point"; then
        log "ERROR" "Repair failed: $mnt_point"
        overall_rc=1
    fi
done

exit "$overall_rc"