From 19d766449eb7ad36602042318f20e77a50d80c94 Mon Sep 17 00:00:00 2001
From: Tomas Kopecek <tkopecek@redhat.com>
Date: Nov 22 2019 12:43:46 +0000
Subject: filesystem pruning cronjobs


---

diff --git a/README.md b/README.md
index c570ec9..6e68947 100644
--- a/README.md
+++ b/README.md
@@ -36,6 +36,10 @@ Supplementary tools and utilities for the Koji build system
 * `koji-list-built-with` - List all the builds that had a given build in
   the buildroot.
 
+* `koji-prune-scratch` - Sample cronjob for cleaning scratch build data
+
+* `koji-prune-work` - Sample cronjob for cleaning work directories
+
 * `koji-remove-arch` - Globally remove an architecture from all tags in
   Koji.
 
diff --git a/src/bin/koji-prune-scratch b/src/bin/koji-prune-scratch
new file mode 100755
index 0000000..ee642dc
--- /dev/null
+++ b/src/bin/koji-prune-scratch
@@ -0,0 +1,21 @@
+#!/bin/bash
+# remove old scratch builds
+
+TOPDIR=/mnt/koji
+
+cd $TOPDIR/scratch/
+
+# we completely remove those that are old enough
+# scratch directories are /mnt/koji/scratch/$username/task_$taskid/
+# note that $username might contain a slash (e.g. host principals)
+find $TOPDIR/scratch/ -mindepth 2 -type d -name 'task_*' -prune -mtime +21 -exec rm -rf {} \;
+
+# For content besides srpms/logs/poms we prune much more aggressively
+# note that this step normally alters the mtime of the task dir, effectively
+# adding to the above retention window.
+for taskdir in $(find $TOPDIR/scratch/ -mindepth 2 -type d -name 'task_*' -prune -mtime +7)
+do
+    find "$taskdir" -type f \! -name '*.src.rpm' \! -name '*.log' \! -name '*.pom' -delete
+done
+
+find $TOPDIR/scratch/ -maxdepth 1 -type d -mtime +1 -empty -delete
diff --git a/src/bin/koji-prune-work b/src/bin/koji-prune-work
new file mode 100755
index 0000000..0dfdb50
--- /dev/null
+++ b/src/bin/koji-prune-work
@@ -0,0 +1,19 @@
+#!/bin/bash
+
+# remove old stuff under kojiroot/work
+
+TOPDIR=/mnt/koji
+TIMEARG="-mtime +2"
+
+# for tasks, try to remove as a unit
+for x in $(find "$TOPDIR"/work/tasks/ -mindepth 2 -maxdepth 2 -type d $TIMEARG); do
+    find "$x" -xdev '!' -type d -print0 | xargs -0 -r rm -f
+    find "$x" -xdev -depth -type d -print0 | xargs -0 -r rmdir
+done
+
+# for anything else, just remove old stuff
+# but don't remove the top level dirs (e.g. cli-build)
+for x in $(find "$TOPDIR"/work -maxdepth 1 -mindepth 1 \! -name tasks); do
+    find "$x" -xdev '!' -type d $TIMEARG -print0 | xargs -0 -r rm -f
+    find "$x" -xdev -depth -mindepth 1 -type d -empty -print0 | xargs -0 -r rmdir
+done