#!/usr/bin/env bash
#
# JbTecWiz Support Centre -- generated fix script
#
# Fault : NFS: "Stale file handle" and hung mounts
# Fix   : Fix a mount that is refused
# Source: https://jbtecwiz.com/support/lnx-sto-nfs-stale
#
# Run as     : Both ends
# Expect     : 25 minutes
# Risk       : low
# Reversible : yes
#
# WHEN THIS IS THE RIGHT FIX
#   mount.nfs reports access denied or connection refused.
#
# HOW TO UNDO IT
#   Nothing persistent changed unless fstab was edited.
#
# Walks the fix one step at a time and asks before each. Steps with no
# command are yours to do -- it prints those and waits. DRYRUN=1 prints
# without executing; UNATTENDED=1 does not ask.
#
# --------------------------------------------------------------------
# NO WARRANTY - USE AT YOUR OWN RISK
#
# This script is provided by JbTecWiz as-is and with no warranty of any
# kind, express or implied. You run it entirely at your own risk.
#
# JbTecWiz accepts no liability for any loss or damage arising from its
# use, including but not limited to data loss, downtime, or configuration
# changes that turn out to be wrong for your system.
#
# You are responsible for reading this script before running it, for
# satisfying yourself that it suits the machine in front of you, and for
# having a working backup first. Some steps cannot be undone.
# --------------------------------------------------------------------

set -uo pipefail
DRYRUN="${DRYRUN:-0}"
UNATTENDED="${UNATTENDED:-0}"
failed=0

rule() { printf "\n%s\n" "$(printf '-%.0s' $(seq 1 70))"; if [ $# -gt 0 ]; then echo "$1"; fi; }
prose() { echo "$1" | fold -s -w 74 | sed "s/^/  /"; }

# Returns 0 when the caller should run the command, 1 when it should not.
# A manual step always returns 1 -- there is nothing for the caller to run.
step() {  # step <n> <do> <why> <manual|cmd> [command lines...]
  local n="$1" dotext="$2" why="$3" mode="$4"; shift 4
  rule "  Step $n of 5"
  prose "$dotext"
  if [ -n "$why" ]; then echo; prose "$why"; fi
  if [ "$mode" = "manual" ]; then
    echo; echo "  -> Do this yourself, then press Enter to carry on."
    if [ "$UNATTENDED" = "0" ] && [ "$DRYRUN" = "0" ]; then read -r _; fi
    return 1
  fi
  echo; printf "    %s\n" "$@"; echo
  if [ "$DRYRUN" = "1" ]; then echo "  (dry run -- not executed)"; return 1; fi
  if [ "$UNATTENDED" = "0" ]; then
    read -r -p "  Run this step? [Y]es / [S]kip / [Q]uit " a
    case "$a" in
      [Qq]*) echo "  Stopped at your request."; exit 0 ;;
      [Ss]*) echo "  Skipped."; return 1 ;;
    esac
  fi
  return 0
}

rule
echo "  NFS: "Stale file handle" and hung mounts"
echo "  Fix a mount that is refused"
echo
echo "  Risk: low   Reversible   25 minutes"
echo
prose 'No warranty. Use at your own risk - JbTecWiz accepts no liability. Read it before you run it, and have a backup.'
rule
echo
if [ "$UNATTENDED" = "0" ] && [ "$DRYRUN" = "0" ]; then
  read -r -p "  Ready? [y/N] " go
  case "$go" in [Yy]*) ;; *) echo "  Nothing was changed."; exit 0;; esac
fi

if step 1 'Check what the server actually exports and to whom.' '' cmd 'showmount -e nfsserver' 'sudo exportfs -v'; then
  showmount -e nfsserver
  sudo exportfs -v
  if [ $? -ne 0 ]; then
    failed=$((failed+1))
    echo "  Step 1 failed. The rest of the fix may depend on it." >&2
  fi
fi

step 2 'Check the client'\''s address matches the export'\''s allowed range -- an export for 10.0.0.0/24 refuses a client that has moved to another subnet.' '' manual || true

if step 3 'Check the firewall. NFSv4 needs only TCP 2049; NFSv3 needs rpcbind and several other ports.' 'NFSv3'\''s dynamic ports are why it is so often blocked by a firewall while NFSv4 works. Forcing version 4 is frequently the whole fix.' cmd 'sudo ss -tlnp | grep 2049' 'rpcinfo -p nfsserver | head'; then
  sudo ss -tlnp | grep 2049
  rpcinfo -p nfsserver | head
  if [ $? -ne 0 ]; then
    failed=$((failed+1))
    echo "  Step 3 failed. The rest of the fix may depend on it." >&2
  fi
fi

if step 4 'Force the version explicitly to remove the negotiation from the equation.' '' cmd 'sudo mount -t nfs -o vers=4.2 nfsserver:/srv/share /mnt/nfs'; then
  sudo mount -t nfs -o vers=4.2 nfsserver:/srv/share /mnt/nfs
  if [ $? -ne 0 ]; then
    failed=$((failed+1))
    echo "  Step 4 failed. The rest of the fix may depend on it." >&2
  fi
fi

if step 5 'Check root squashing if files are visible but not writable -- root on the client becomes nobody on the server by default.' '' cmd 'sudo grep -n squash /etc/exports' 'id; ls -ln /mnt/nfs | head'; then
  sudo grep -n squash /etc/exports
  id; ls -ln /mnt/nfs | head
  if [ $? -ne 0 ]; then
    failed=$((failed+1))
    echo "  Step 5 failed. The rest of the fix may depend on it." >&2
  fi
fi

rule "  Confirm it worked"
prose 'The mount succeeds and a test file can be created and removed.'
if [ "$DRYRUN" = "0" ]; then
  sudo touch /mnt/nfs/.writetest && sudo rm /mnt/nfs/.writetest && echo writable
fi

rule
if [ "$failed" -gt 0 ]; then
  echo "  Finished with $failed failed step(s)."
  echo "  Read the full write-up at https://jbtecwiz.com/support/lnx-sto-nfs-stale"
else
  echo "  Finished."
fi
echo
prose 'To undo: Nothing persistent changed unless fstab was edited.'
rule