do not remove local bk if upload failed

README.md

@@ -197,6 +197,12 @@ variables:
     # take selective backup
     EXTRA_EXCLUDES="/jails/test.dom.com" /usr/local/sbin/zfsbk.sh mybk 1
 
+`zfsbk.sh` maintains serialized backups in directory `/backups` after having uploaded them.
+If you want to cleanup the backup file locally after uploading it upstream, set the
+`CLEAR_BKFILE` environment variable to a non-empty value. `zfsbk.sh` will maintain the local
+backup regardless of this variable if the upload failed.
+
+
 # Managing multiple ZFS Pools
 
 Simply run these scripts once for each pool, setting the _ZPOOL_ environment variable
@@ -215,7 +221,7 @@ tested for a number of scenarios, and used in a number of servers.
 Because of the inherently partial nature of testing, you are aware that this
 brings no guarantees that it'll work for your scenario.
 
-Careful yourself when working with ZFS snapshots! Check this out:
+**Careful yourself** when working with ZFS snapshots! Check this out:
 
     zfs destroy -r zroot@mysnap     # destroy snapshot 'mysnap' from pool 'zroot'
     zfs destroy -r zroot#mysnap     # destroy all system data, and comment it's 'mysnap'

zfsbk.sh

@@ -114,11 +114,11 @@ then
         echo "Expecting rsync://.. or scp://.."
         exit 1
     fi
-fi
-
-# remove backup file if requested
-if [ "x$CLEAR_BKFILE" != x ]
-then
-    rm -f $bkfile
+    # remove local backup if requested & upload was successful
+    if [ $? -eq 0 -a "x$CLEAR_BKFILE" != x ]
+    then
+        # remove backup file if requested
+        rm -f $bkfile
+    fi
 fi