eproducible: support for saving artifacts from the builds. use 'artifacts' as the first package when manual scheduling and packages will save files!

4 files changed, 38 insertions, 4 deletions

diff --git a/bin/reproducible_build.sh b/bin/reproducible_build.sh
index d76e52ed..e281e896 100755
--- a/bin/reproducible_build.sh
+++ b/bin/reproducible_build.sh
@@ -24,6 +24,18 @@ create_results_dirs() {
 }
 
 cleanup_all() {
+	if [ "$SAVE_ARTIFACTS" == 1 ] ; then
+		local hash=$(tr -cd '[:alnum:]' < /dev/urandom | fold -w5 | head -1)
+		local ARTIFACTS="artifacts/r00t-me/tmp-${hash}/${SUITE}/${PACKAGE}"
+		mkdir -p "/var/lib/jenkins/userContent/$ARTIFACTS"
+		cp -r $TMPDIR/* "/var/lib/jenkins/userContent/$ARTIFACTS"
+		echo | tee -a ${RBUILDLOG}
+		echo "Artifacts from this build are preserved. They will be available for very short time so download them now if you want them." | tee -a ${RBUILDLOG}
+		echo "You shouldn't trust packages you downloaded from this host, they can contain malware or the worst of your fear, packaged in a debian format." | tee -a ${RBUILDLOG}
+		echo "If you are not afraid of facing your fears helping the world by investigating reproducible build issues, you can download the artifacts from the following url:" | tee -a ${RBUILDLOG}
+		echo "https://reproducible.debian.net/$ARTIFACTS" | tee -a ${RBUILDLOG}
+		echo | tee -a ${RBUILDLOG}
+	fi
 	rm -r $TMPDIR $TMPCFG
 }
 
@@ -141,7 +153,7 @@ for i in $SUITES ; do
 done
 SQL_SUITES="$SQL_SUITES)"
 
-RESULT=$(sqlite3 -init $INIT ${PACKAGES_DB} "SELECT s.suite, s.id, s.name, sch.date_scheduled FROM schedule AS sch JOIN sources AS s ON sch.package_id=s.id WHERE sch.date_build_started = '' AND s.suite IN $SQL_SUITES ORDER BY date_scheduled LIMIT 1")
+RESULT=$(sqlite3 -init $INIT ${PACKAGES_DB} "SELECT s.suite, s.id, s.name, sch.date_scheduled, sch.save_artifacts FROM schedule AS sch JOIN sources AS s ON sch.package_id=s.id WHERE sch.date_build_started = '' AND s.suite IN $SQL_SUITES ORDER BY date_scheduled LIMIT 1")
 if [ -z "$RESULT" ] ; then
 	echo "No packages scheduled, sleeping 30m."
 	sleep 30m
@@ -151,6 +163,7 @@ else
 	SRCPKGID=$(echo $RESULT|cut -d "|" -f2)
 	SRCPACKAGE=$(echo $RESULT|cut -d "|" -f3)
 	SCHEDULED_DATE=$(echo $RESULT|cut -d "|" -f4)
+	SAVE_ARTIFACTS=$(echo $RESULT|cut -d "|" -f5)
 	create_results_dirs
 	echo "============================================================================="
 	echo "Trying to reproducibly build ${SRCPACKAGE} in ${SUITE} on ${ARCH} now."
diff --git a/bin/reproducible_common.sh b/bin/reproducible_common.sh
index f428d6ad..7f318e14 100755
--- a/bin/reproducible_common.sh
+++ b/bin/reproducible_common.sh
@@ -79,8 +79,10 @@ schedule_packages() {
 	# so schedule them in the past, so they are picked earlier :)
 	DATE="2014-10-01 00:23"
 	TMPFILE=$(mktemp)
+	ARTIFACTS=$1
+	shift
 	for PKG_ID in $@ ; do
-		echo "REPLACE INTO schedule (package_id, date_scheduled, date_build_started) VALUES ('$PKG_ID', '$DATE', '');" >> $TMPFILE
+		echo "REPLACE INTO schedule (package_id, date_scheduled, date_build_started, save_artifacts) VALUES ('$PKG_ID', '$DATE', '', '$ARTIFACTS');" >> $TMPFILE
 	done
 	cat $TMPFILE | sqlite3 -init $INIT ${PACKAGES_DB}
 	rm $TMPFILE
diff --git a/bin/reproducible_maintainance.sh b/bin/reproducible_maintainance.sh
index 84624bb6..97e588d2 100755
--- a/bin/reproducible_maintainance.sh
+++ b/bin/reproducible_maintainance.sh
@@ -101,7 +101,8 @@ if [ ! -z "$FAILED_BUILDS" ] ; then
 		check_candidates
 		if [ $TOTAL -ne 0 ] ; then
 			echo " - in $SUITE: $CANDIDATES"
-			schedule_packages $PACKAGE_IDS
+			# '0' here means the artifacts will not be saved
+			schedule_packages 0 $PACKAGE_IDS
 		fi
 	done
 	DIRTY=true
@@ -200,6 +201,15 @@ if [ ! -z "$OLDSTUFF" ] ; then
 	echo
 fi
 
+# remove artifacts older than 3 days
+mkdir -p /var/lib/jenkins/userContent/artifacts
+ARTIFACTS=$(find /var/lib/jenkins/userContent/artifacts -maxdepth 1 -type d -mtime +3 -exec rm -rv {} \;)
+if [ ! -z "$ARTIFACTS" ] ; then
+	echo
+	echo "Removed $ARTIFACTS artifacts."
+	echo
+fi
+
 if ! $DIRTY ; then
 	echo "Everything seems to be fine."
 	echo
diff --git a/bin/reproducible_schedule_on_demand.sh b/bin/reproducible_schedule_on_demand.sh
index 74321e26..c4bd9877 100755
--- a/bin/reproducible_schedule_on_demand.sh
+++ b/bin/reproducible_schedule_on_demand.sh
@@ -22,6 +22,15 @@ if [ "$SUITE" = "sid" ] ; then
 	SUITE=unstable
 fi
 
+ARTIFACTS=0
+if [ $1 = "artifacts" ] ; then
+	ARTIFACTS=1
+	shift
+	echo
+	echo "The artifacts of these builds will be saved. Look at the build log for the link"
+	echo
+fi
+
 CANDIDATES="$@"
 check_candidates
 if [ ${#PACKAGE_IDS} -gt 256 ] ; then
@@ -34,7 +43,7 @@ fi
 MESSAGE="$TOTAL $PACKAGES_TXT $ACTION in $SUITE: ${PACKAGES_NAMES:0:256}$BLABLABLA"
 
 # finally
-schedule_packages $PACKAGE_IDS
+schedule_packages $ARTIFACTS $PACKAGE_IDS
 echo
 echo "$MESSAGE"
 if [ -z "${BUILD_URL:-}" ] && [ $TOTAL -ne 0 ] ; then