com.sun:auto-snapshot support, remove install scripts, add narrow_recursive_filesystems and auto_include

src/lib/svc/method/zfs-auto-snapshot

@@ -65,7 +65,7 @@ SEP=":"
 
 # This variable gets set to the restarter/logfile property
 # whenever we have $FMRI defined. Used by the print_log and
-# print_note functions below for all output, it's definied
+# print_note functions below for all output, it's defined
 # by the schedule_snapshots take_snapshots and unschedule_snapshots
 # methods.
 LOG=""
@@ -90,7 +90,7 @@ function zfs_smf_props {
 
 IFS="
 "
-SMF_PROPS="$(svcprop -t -p zfs $1 |\
+SMF_PROPS="$(svcprop -t -p zfs -p restarter/logfile $1 |\
             sed -e 's#zfs/fs-name#zfs/fs_name#g' \
         -e 's#zfs/backup-lock#zfs/backup_lock#g' \
         -e 's#zfs/snapshot-children#zfs/snapshot_children#g' \
@@ -113,6 +113,9 @@ function schedule_snapshots {
 
 	typeset FMRI=$1
 	zfs_smf_props $FMRI 
+	# functions we call need $LOG set.
+	export LOG=$logfile
+
 	# FIXME need work in here to actually validate the FMRI props
 	typeset FILESYS="$fs_name"
 	typeset INTERVAL="$interval"
@@ -136,9 +139,7 @@ function schedule_snapshots {
 	typeset OFFSET=0
 
 	case $FILESYS in
-	"//")
+	'//' | '##')
-		;;
-	"##" )
 		;;
 	*)
 		# validate the filesystem
@@ -376,8 +377,8 @@ function take_snapshot {
 	FMRI=$1
 	zfs_smf_props $FMRI
 
-	export LOG=$log
+	export LOG=$logfile
-	typeset DATE=$(date +%F-%H${SEP}%M${SEP}%S)
+	typeset DATE=$(date +%F-%H${SEP}%M)
 	typeset FILESYS="$fs_name"
 	typeset KEEP=$keep
 	typeset SNAP_CHILDREN=$snapshot_children
@@ -532,7 +533,7 @@ function destroy_older_snapshots {
 	        print_note "$snapshot being destroyed ${RECURSIVE} as per \
  retention policy."
 	        zfs destroy ${RECURSIVE} $snapshot
-	        check_failure $? "Unable to destroy $snapshot"
+	        check_failure $? "Unable to destroy $snapshot" "NON_FATAL"
 	    else
 	        # don't destroy this one			
 	        COUNTER=$(($COUNTER - 1))
@@ -541,20 +542,25 @@ function destroy_older_snapshots {
 }
 
 # Given the exit status of a command, an integer, 0 if the command completed
-# without errors. If the command exited with errors, we degrade the
+# without errors. If the command exited with errors we degrade the
-# state of this service into maintenance mode. We also log an error message
+# state of this service into maintenance mode. If a 3rd argument is presented
-# as passed into this function.
+# we don't degrade the service. We also log an error message as passed into
+# this function.
 #
-function check_failure { # integer exit status, error message to display
+function check_failure { # integer exit status, error message to display, be fatal
 
 	typeset RESULT=$1
 	typeset ERR_MSG=$2
+	typeset NON_FATAL=$3
 
 	if [ $RESULT -ne 0 ] ; then
 	    print_log "Error: $ERR_MSG"
+	    print_log "Moving service $FMRI to maintenance mode."
+	    if [ -z "${NON_FATAL}" ] ; then
 		print_log "Moving service $FMRI to maintenance mode."
 	        svcadm mark maintenance $FMRI
 	    fi
+	fi
 
 }
 
@@ -716,28 +722,58 @@ function get_snapshot_datasets { #LABEL #SNAP_CHILDREN
 	typeset LABEL=$1
 	typeset SNAP_CHILDREN=$2
 	if [ "${SNAP_CHILDREN}" = "true" ] ; then
-	    typeset FS=$(zfs get -s local -o name,value com.sun:auto-snapshot:$LABEL \
+		# FIXME this doesn't cope with the case where a dataset is
-	        | grep true$ | awk '{print $1}')
+		# set locally to differing values - value for :LABEL needs to override
+		typeset FS=$(zfs get -H -s local \
+		-o name,value com.sun:auto-snapshot,com.sun:auto-snapshot:$LABEL \
+		| grep -v 'false$' | egrep ' true '\|'true$' | awk '{print $1}' | sort)
+		FS="$(narrow_recursive_filesystems $FS)"
 	else
 		typeset FS=$(zfs list -t filesystem,volume \
-		-o name,com.sun:auto-snapshot:$LABEL \
+		-o name,com.sun:auto-snapshot,com.sun:auto-snapshot:$LABEL \
-	        | grep true$ | awk '{print $1}')
+		| grep -v 'false$' | egrep ' true '\|'true$' | awk '{print $1}')
 	fi
 	echo "$FS"
 }
 
 # Get a list of filesystems we should snapshot. We look for all filesystems
 # and volumes that don't have a property com.sun:auto-snapshot:$LABEL
-# set to false.
+# or com.sun:auto-snapshot set to false. The :$LABEL specifier overrides
+# the more general property setting.
 function get_nonexcluded_datasets {
 
 	typeset LABEL=$1
 	typeset FS=$(zfs list -H -t filesystem,volume \
-	    -o name,com.sun:auto-snapshot:$LABEL \
+	    -o name,com.sun:auto-snapshot,com.sun:auto-snapshot:$LABEL \
-	    | grep -v false$ | awk '{print $1}')
+	    | egrep -v '	false	-$'\|'false$' | awk '{print $1}')
 	echo "$FS"
 }
 
+# Given a sorted list of filesystems, determine whether any of the
+# listed filesystems are redundant
+# eg. for tank/other tank/foo tank/other/bar tank/foo/bar
+# we only need to snapshot tank/other and tank/foo
+function narrow_recursive_filesystems {
+	# for each filesystem in the list, get each of it's ancestors
+	# if any of the ancestors is already in the list, don't add it,
+	# otherwise, do.
+	typeset LIST=""
+	for ds in $@ ; do
+		ANCESTOR_IN_LIST=""
+		ancestor=$(dirname $ds)
+		while [ $ancestor != "." ] ; do
+			if echo $LIST | fgrep $ancestor// > /dev/null ; then
+				ANCESTOR_IN_LIST=true
+			fi
+			ancestor=$(dirname $ancestor)
+		done
+		if [ -z "${ANCESTOR_IN_LIST}" ] ; then
+			LIST="${LIST} ${ds}//"
+		fi
+	done
+	echo ${LIST} | sed -e 's#//##g'
+}
+
 
 
 # Determine if a pool is currently being scrubbed or resilvered.
@@ -745,7 +781,7 @@ function get_nonexcluded_datasets {
 
 # The 2nd arg is a cache of pools known to be not scrubbing during this
 # invocation of the script. This does risk a scrub starting mid-way through
-# the script being started and us not checking for it - but if that's just
+# the script and us not checking for it - but if that's just
 # happened, then restarting the scrub as a result of a snapshot being taken
 # won't be too expensive.
 function is_scrubbing { # POOL SCRUBLIST
@@ -770,6 +806,33 @@ function is_scrubbing { # POOL SCRUBLIST
 	fi
 }
 
+# This function runs on startup - by default, if we're taking snapshots 
+# under a // schedule, and there isn't a property set on the pool
+# com.sun:auto-snapshot=false, then we set the property to true, causing
+# all datasets on the system to get included by the service. 
+function auto_include {
+	FS_NAME=$fs_name
+	LABEL=$label
+	if [ "$FS_NAME" == "//" ] ; then
+		POOLS=$(zpool list -H -o name)
+		for pool in $POOLS ; do
+			if ! zpool status -x $pool | grep "state: UNAVAIL" > /dev/null ; then
+				SNAPALL=$(zfs get com.sun:auto-snapshot $pool)
+				SNAPLABEL=$(zfs get com.sun:auto-snapshot:$LABEL $pool)
+				SNAP=$SNAPALL$SNAPLABEL
+				case $SNAP in
+				*true | true*)
+					;;
+				*false | false*)
+					;;
+				*)
+					zfs set com.sun:auto-snapshot=true $pool
+					;;
+				esac
+			fi
+		done	
+	fi
+}
 
 
 
@@ -784,12 +847,13 @@ function is_scrubbing { # POOL SCRUBLIST
 
 if [ -n "${SMF_FMRI}" ] ; then
 	zfs_smf_props $SMF_FMRI
-	export LOG=${log}
+	export LOG=$logfile
 fi
 
-# $1 start | stop | an FMRI that we want to take snapshots of.
+# $1 start | stop | refresh | an FMRI that we want to take snapshots of.
 case "$1" in
 'start')
+	auto_include $SMF_FMRI
 	schedule_snapshots $SMF_FMRI
 	if [ $? -eq 0 ] ; then
 	    result=$SMF_EXIT_OK
@@ -808,7 +872,6 @@ case "$1" in
 	    result=$SMF_EXIT_ERR_FATAL
 	fi
         ;;
-
 # the default case, we actually call from the cron job itself that's
 # executing this script, and do the job of taking snapshots.
 *)

src/pkginfo

@@ -1,4 +1,3 @@
-CLASSES=none 
 LC_MESSAGES=en_US.UTF-8
 LANG=en_US.UTF-8
 TZ=Eire
@@ -9,13 +8,14 @@ PKG=TIMFauto-snapshot
 NAME=ZFS Automatic Snapshot Service
 ARCH=all
 BASEDIR=/
-VERSION=0.11
+VERSION=0.11ea
 MAXINST=1
 CATEGORY=application
 DESC=Takes automatic snapshots of ZFS filesystems on a periodic basis.
-PSTAMP=haiiro20080730110036
+PSTAMP=haiiro20080822102134
 VENDOR=Sun Microsystems, Inc.
 HOTLINE=Please contact your local service provider
+CLASSES=none manifest
 EMAIL=tim.foster@sun.com
 SUNW_PKGVERS=1.0
 SUNW_PKG_ALLZONES=false

src/postinstall

@@ -27,22 +27,4 @@
 
 # a postinstall script - it should import the manifests
 # and enable the service
-DEFAULT=svc:/system/filesystem/zfs/auto-snapshot:default
+echo "Post install script doing nathing!"
-
-FMRIS="svc:/system/filesystem/zfs/auto-snapshot:frequent svc:/system/filesystem/zfs/auto-snapshot:hourly svc:/system/filesystem/zfs/auto-snapshot:daily svc:/system/filesystem/zfs/auto-snapshot:weekly svc:/system/filesystem/zfs/auto-snapshot:monthly"
-
-FILES="auto-snapshot-daily.xml     auto-snapshot-monthly.xml auto-snapshot-frequent.xml  auto-snapshot-weekly.xml auto-snapshot-hourly.xml    zfs-auto-snapshot.xml"
-
-/usr/sbin/svccfg import /var/svc/manifest/system/filesystem/zfs-auto-snapshot.xml
-
-for manifest in $FILES
-do
-	echo Importing $manifest
-	/usr/sbin/svccfg import /var/svc/manifest/system/filesystem/$manifest
-done
-
-for fmri in $FMRIS
-do
-	echo Enabling $fmri
-	/usr/sbin/svcadm enable $fmri
-done

src/preremove

@@ -27,14 +27,4 @@
 
 # a postinstall script - it should disable the services
 # and delete the instances
-DEFAULT=svc:/system/filesystem/zfs/auto-snapshot:default
+echo preremove script doing nothing
-
-FMRIS="svc:/system/filesystem/zfs/auto-snapshot:frequent svc:/system/filesystem/zfs/auto-snapshot:hourly svc:/system/filesystem/zfs/auto-snapshot:daily svc:/system/filesystem/zfs/auto-snapshot:weekly svc:/system/filesystem/zfs/auto-snapshot:monthly"
-
-for fmri in $FMRIS $DEFAULT ; do
-	STATE=$(/usr/bin/svcs -H -o state $fmri)
-	if [ "$STATE" = "online" ] ; then
-	    /usr/sbin/svcadm disable -s $fmri
-	fi
-	/usr/sbin/svccfg delete $fmri
-done

src/prototype

@@ -2,24 +2,26 @@ i pkginfo
 i copyright
 i postinstall
 i preremove
+i i.manifest
+i r.manifest
+d none lib 0755 root bin 
+d none lib/svc 0755 root bin 
+d none lib/svc/method 0755 root bin 
+f none lib/svc/method/zfs-auto-snapshot 0755 root bin 
 d none var 0755 root sys	
 d none var/svc 0755 root sys
 d none var/svc/manifest 0755 root sys
 d none var/svc/manifest/system 0755 root sys
 d none var/svc/manifest/system/filesystem 0755 root sys
-f none var/svc/manifest/system/filesystem/zfs-auto-snapshot.xml 0644 root sys
+f manifest var/svc/manifest/system/filesystem/zfs-auto-snapshot.xml 0644 root sys
-f none var/svc/manifest/system/filesystem/auto-snapshot-monthly.xml 0644 root sys	
+f manifest var/svc/manifest/system/filesystem/auto-snapshot-monthly.xml 0644 root sys	
-f none var/svc/manifest/system/filesystem/auto-snapshot-frequent.xml 0644 root sys
+f manifest var/svc/manifest/system/filesystem/auto-snapshot-frequent.xml 0644 root sys
-f none var/svc/manifest/system/filesystem/auto-snapshot-daily.xml 0644 root sys
+f manifest var/svc/manifest/system/filesystem/auto-snapshot-daily.xml 0644 root sys
-f none var/svc/manifest/system/filesystem/auto-snapshot-hourly.xml 0644 root sys
+f manifest var/svc/manifest/system/filesystem/auto-snapshot-hourly.xml 0644 root sys
-f none var/svc/manifest/system/filesystem/auto-snapshot-weekly.xml 0644 root sys
+f manifest var/svc/manifest/system/filesystem/auto-snapshot-weekly.xml 0644 root sys
 d none usr 0755 root sys
 d none usr/share 0755 root sys
 d none usr/share/applications 0755 root other
 f none usr/share/applications/automatic-snapshot.desktop 0644 root bin
 d none usr/bin 0755 root bin
 f none usr/bin/zfs-auto-snapshot-admin.sh 0755 root sys
-d none lib 0755 root bin
-d none lib/svc 0755 root bin
-d none lib/svc/method 0755 root bin
-f none lib/svc/method/zfs-auto-snapshot 0755 root bin

src/var/svc/manifest/system/filesystem/auto-snapshot-daily.xml

@@ -62,6 +62,9 @@ snapshots into the past.
 	  <propval name="verbose" type="boolean" value="false"
 		   override="true"/>
 
+	  <propval name="avoidscrub" type="boolean" value="false"
+		   override="false"/>
+
 	</property_group>
 
 	</instance>

src/var/svc/manifest/system/filesystem/auto-snapshot-frequent.xml

@@ -62,6 +62,9 @@ com.sun:auto-snapshot:frequent=true every
 	  <propval name="verbose" type="boolean" value="false"
 		   override="true"/>
 
+          <propval name="avoidscrub" type="boolean" value="false"
+                   override="false"/>
+
 	</property_group>
 
 	</instance>

src/var/svc/manifest/system/filesystem/auto-snapshot-hourly.xml

@@ -63,7 +63,7 @@ and keeps 24 of these snapshots into the past.
 		   override="true"/>
 
 	  <propval name="avoidscrub" type="boolean" value="true"
-		   override="true"/>
+		   override="false"/>
 
 	</property_group>
 

src/var/svc/manifest/system/filesystem/auto-snapshot-monthly.xml

@@ -63,7 +63,7 @@ and keeps 12 of these snapshots into the past.
 		   override="true"/>
 
 	  <propval name="avoidscrub" type="boolean" value="true"
-		   override="true"/>
+		   override="false"/>
 
 	</property_group>
 

src/var/svc/manifest/system/filesystem/auto-snapshot-weekly.xml

@@ -63,7 +63,7 @@ and keeps 4 of these snapshots into the past.
 		   override="true"/>
 
 	  <propval name="avoidscrub" type="boolean" value="true"
-		   override="true"/>
+		   override="false"/>
 
 	</property_group>
 

src/var/svc/manifest/system/filesystem/zfs-auto-snapshot.xml

@@ -59,12 +59,29 @@
 <!-- the properties we expect that any instance will define
 they being :
 	fs-name : The name of the filesystem we want to snapshot.
+
 	The special filesystem name "//" indicates we should
-	look at the com.sun:auto-snapshot:<label> ZFS user
+	look at the com.sun:auto-snapshot ZFS user
-	property on datasets, set to "true" if the dataset
+	properties on datasets, set to "true" if the dataset
 	should have snapshots taken by this instance.
 
-	interval : minutes | hours | days | months
+	The special filesystem name "##" indicates we should
+	snapshot all filesystems on the machine, except those
+	marked with the com.sun:auto-snapshot ZFS user property
+	set to false. The snapshot-children setting is ignored when
+	using this fs-name value.
+
+	For both of the above, setting com.sun:auto-snapshot:<label>
+	will override the general setting for com.sun:auto-snapshot.
+	
+
+	interval : minutes | hours | days | months | none
+
+	For the interval "none" a cron job is not created for that
+	instance - instead the user can manually file the method
+	script to take snapshots defined by the rest of the properties
+	in the instance. The period and offset values are ignored in
+	this case.
 
 	period : How many (m,h,d,m) do we wait between snapshots
 
@@ -101,7 +118,7 @@ they being :
 		in the service printing more detail in the log
 		about what it's doing.
 
-	avoidscrub : Set to true by default, this determines whether
+	avoidscrub : Set to false by default, this determines whether
 		we should avoid taking snapshots on any pools that have
 		a scrub or resilver in progress.
 		More info in the bugid:
@@ -123,7 +140,7 @@ they being :
 
 	  <propval name="label" type="astring" value="" override="true"/>
 	  <propval name="verbose" type="boolean" value="false" override="true"/>
-	  <propval name="avoidscrub" type="boolean" value="true" override="true"/>
+	  <propval name="avoidscrub" type="boolean" value="false" override="true"/>
 
 	</property_group>