Added fatal_and_exit routine to handle all early exits out of the script

diff --git a/zfs-replicate b/zfs-replicate
index 09cb3babb7600f0526a14b4b2c2f945f41660ea4..413d078e05cf28311896be3d7d94bcca454aa6d0 100755 (executable)
--- a/zfs-replicate
+++ b/zfs-replicate
@@ -5,12 +5,14 @@ remote=$1
 remote_fs=$2
 remote_pool=${2%%/*}
 remote_lockdir="/tmp/zfs-admin-lock"
+hostname=`hostname`
 
 # Set the name of the local pool used to store the backup of the remote
 local_pool=backups
 
 # Set the email address to send notification to
 mailto=root@pippins.net
+mailx=/usr/bin/mailx
 
 # When this variable is set, local filesystems will be destroyed 
 # before receiving a full streams into them from the remote source.
@@ -21,23 +23,38 @@ zfs=/usr/sbin/zfs
 
 # Setup our cleanup and exit trap
 cleanup() {
-  ssh $remote rm -rf "$remote_lockdir"
-  exit
+  if [[ -e "$local_list" ]]; then
+    rm -f $local_list
+  fi
+  if [[ -e "$remote_list" ]]; then
+    rm -f $remote_list
+  fi
+  if [[ -n "$remote" ]]; then
+    ssh $remote ls -d "$remote_lockdir" > /dev/null 2>&1
+    if [[ $? == 0 ]]; then
+      ssh $remote rm -rf "$remote_lockdir"
+    fi
+  fi
 }
-trap cleanup INT
+fatal_and_exit() {
+  echo -e 2>&1 "$1"
+  if [[ -n "$2" ]]; then
+    echo -e "$1" | $mailx -s "zfs replicate on $hostname failed" "$2"
+  fi
+  exit 1
+}
+trap fatal_and_exit INT
 trap cleanup EXIT
 
 # Make sure we have valid arguments
 if [[ -z "$remote" ]] || [[ -z "$remote_fs" ]]; then
-  echo "Usage: $0 <hostname> <zfs filesystem>"
-  exit 1
+  fatal_and_exit "Usage: $0 <hostname> <zfs filesystem>"
 fi
 
 # Make sure the local pool and local receiving filesystem exist, or print some errors
 zpool list -H "$local_pool" >/dev/null 2>&1
 if [ $? != 0 ]; then
-  echo >&2 "-E- The local pool, '$local_pool' doesn't seem to exist."
-  exit 1
+  fatal_and_exit "-E- The local pool, '$local_pool' doesn't seem to exist." $mailto
 fi
 zfs list "$local_pool/$remote_pool" >/dev/null 2>&1
 if [ $? != 0 ]; then
@@ -45,8 +62,7 @@ if [ $? != 0 ]; then
   echo >&2 "    Creating the local filesystem to receive the remote pool into: $local_pool/$remote_pool"
   $zfs create $local_pool/$remote_pool
   if [ $? != 0 ]; then
-    echo "-E- remote $zfs create command failed"
-    exit 1
+    fatal_and_exit "-E- remote $zfs create command failed" $mailto
   fi
 fi
 
@@ -54,8 +70,7 @@ fi
 local_pool_guid=`zpool get guid $local_pool 2>&1 | grep $local_pool | awk '{ print $3 }'`
 zpool get guid $local_pool > /dev/null 2>&1
 if [ $? != 0 ]; then
-  echo >&2 "-E- Unable to extract the guid for the local pool: $local_pool"
-  exit 1
+  fatal_and_exit "-E- Unable to extract the guid for the local pool: $local_pool" $mailto
 fi
 
 # Turn on shell verbosity
@@ -80,9 +95,9 @@ while true; do
   fi
   if [[ $attempts -gt $maxattempts ]]; then
     # We've exceeded our maximum while loop count
-    echo "-W- The zfs filesystem has been locked down. Skipping replicate operation."
-    ssh $remote ls -ld $remote_lockdir | /usr/bin/mailx -s "zfs-replicate-all unable to obtain zfs admin lock" $mailto
-    exit 1
+    echo "-E- The zfs filesystem has been locked down. Skipping replicate operation."
+    fail_msg=`ssh $remote ls -ld $remote_lockdir 2>&1`
+    fatal_and_exit "zfs-replicate-all unable to obtain zfs admin lock:\n$fail_msg" $mailto
   fi
 done
 
@@ -97,8 +112,7 @@ ssh $remote \
     grep ^${remote_fs}@ |
     awk '{print$1}' > $remote_list
 if [ $? != 0 ]; then
-  echo "-E- remote $zfs list command failed"
-  exit 1
+  fatal_and_exit "-E- remote $zfs list command failed" $mailto
 fi
 
 # List the snapshots on the local machine.
@@ -110,8 +124,7 @@ $zfs list -H -t snapshot |
     grep -v ^${local_pool}/${current_backup_marker} | 
     awk "{gsub(/^$local_pool./,\"\",\$1); print\$1}" > $local_list
 if [ $? != 0 ]; then
-  echo "-E- local $zfs list command failed"
-  exit 1
+  fatal_and_exit "-E- local $zfs list command failed" $mailto
 fi
 
 # Destroy the current backup marker snapshot on the remote system if it exists
@@ -119,16 +132,14 @@ grep -q ${current_backup_marker} $remote_list
 if [ $? == 0 ]; then
   ssh $remote $zfs destroy ${current_backup_marker} 
   if [ $? != 0 ]; then
-    echo "-E- remote $zfs destroy command failed"
-    exit 1
+    fatal_and_exit "-E- remote $zfs destroy command failed" $mailto
   fi
 fi
 
 # Create the current backup marker snapshot on the remote system
 ssh $remote $zfs snapshot ${current_backup_marker}
 if [ $? != 0 ]; then
-  echo "-E- remote $zfs snapshot command failed"
-  exit 1
+  fatal_and_exit "-E- remote $zfs snapshot command failed" $mailto
 fi
 
 # Check to see if the previous backup marker exists in the remote snapshot list.
@@ -150,15 +161,13 @@ if [ $no_markers == 0 ]; then
   # previous backup marker again from a previously interrupted replicate.
   $zfs rollback -r ${local_pool}/${previous_backup_marker} 
   if [ $? != 0 ]; then
-    echo "-E- remote incremental $zfs rollback command failed"
-    exit 1
+    fatal_and_exit "-E- remote incremental $zfs rollback command failed" $mailto
   fi
   # Now it should be safe to send the snaps
   ssh $remote $zfs send -R -I${previous_backup_marker} ${current_backup_marker} | 
       $zfs receive -vF -d ${local_pool}/${remote_pool}
   if [ $? != 0 ]; then
-    echo "-E- remote incremental $zfs send command failed"
-    exit 1
+    fatal_and_exit "-E- remote incremental $zfs send command failed" $mailto
   fi
 else
   # We didn't find any backup markers, next check to see if we have a common snapshot.
@@ -178,8 +187,7 @@ else
     ssh $remote $zfs send -R -I${common/*@/@} ${current_backup_marker} |
         $zfs receive -vF -d ${local_pool}/${remote_pool}
     if [ $? != 0 ]; then
-      echo "-E- remote incremental $zfs send command failed"
-      exit 1
+      fatal_and_exit "-E- remote incremental $zfs send command failed" $mailto
     fi
   else
     # We did not find any markers or a common snapshot
@@ -190,22 +198,19 @@ else
       if [[ $destroy_local_filesystem_on_full_replicate == 1 ]]; then
         $zfs destroy -r ${local_pool}/${remote_fs}
         if [ $? != 0 ]; then
-          echo "-E- remote full $zfs destroy command failed"
-          exit 1
+          fatal_and_exit "-E- remote full $zfs destroy command failed" $mailto
         fi
       else
         echo "-W- We need to destroy a local filesystem before receiving a full stream."
        echo "    However, since the option is set to prevent this, skipping replicate operation."
-        echo "$zfs destroy -r ${local_pool}/${remote_fs} not able to run" | /usr/bin/mailx -s "zfs-replicate-all unable to destroy local filesystem" $mailto
-        exit 1
+        fatal_and_exit "unable to destroy local filesystem:\n$zfs destroy -r ${local_pool}/${remote_fs} not able to run" $mailto
       fi
     fi
     # Send the full filesystem
     ssh $remote $zfs send -R ${current_backup_marker} |
         $zfs receive -vF -d ${local_pool}/${remote_pool}
     if [ $? != 0 ]; then
-      echo "-E- remote full $zfs send command failed"
-      exit 1
+      fatal_and_exit "-E- remote full $zfs send command failed" $mailto
     fi
   fi
 fi
@@ -218,15 +223,10 @@ ssh $remote $zfs destroy ${previous_backup_marker} > /dev/null 2>&1
 # Rename the current backup marker to be the previous backup marker
 $zfs rename ${local_pool}/${current_backup_marker} ${local_pool}/${previous_backup_marker}
 if [ $? != 0 ]; then
-  echo "-E- local $zfs rename command failed"
-  exit 1
+  fatal_and_exit "-E- local $zfs rename command failed" $mailto
 fi
 ssh $remote $zfs rename ${current_backup_marker} ${previous_backup_marker}
 if [ $? != 0 ]; then
-  echo "-E- remote $zfs rename command failed"
-  exit 1
+  fatal_and_exit "-E- remote $zfs rename command failed" $mailto
 fi
-   
-# Remove tmp files
-rm -f $local_list $remote_list