From 003303a0658832a3b54797fa340e875a30883b26 Mon Sep 17 00:00:00 2001
From: Silvie Chlupova <sisi.chlupova@gmail.com>
Date: Fri, 11 Sep 2020 14:24:39 +0200
Subject: [PATCH] Garbage collection script

This makes subsequent git clones from the dist-git machine much more IO effective and faster.
It also saves disk space. It cleans up all unnecessary files.
The dist-git-gc script will run monthly.
---
 configs/dist-git/dist-git.conf      |  3 +++
 configs/systemd/dist-git-gc.service |  9 +++++++++
 configs/systemd/dist-git-gc.timer   |  8 ++++++++
 dist-git.spec                       |  3 +++
 scripts/dist-git/dist-git-gc        | 11 +++++++++++
 5 files changed, 34 insertions(+)
 create mode 100644 configs/systemd/dist-git-gc.service
 create mode 100644 configs/systemd/dist-git-gc.timer
 create mode 100755 scripts/dist-git/dist-git-gc

diff --git a/configs/dist-git/dist-git.conf b/configs/dist-git/dist-git.conf
index 3667618..c5b9adf 100644
--- a/configs/dist-git/dist-git.conf
+++ b/configs/dist-git/dist-git.conf
@@ -7,6 +7,9 @@ git_author_email = rel-eng@lists.fedoraproject.org
 
 lookaside_dir  = /var/lib/dist-git/cache/lookaside/pkgs
 gitroot_dir    = /var/lib/dist-git/git
+# relates to the dist-git-gc script (and timer),
+# we need to dive 2 levels from gitroot_dir in the directory structure
+git_gc_depth   = 2
 
 gitolite       = True
 grok           = True
diff --git a/configs/systemd/dist-git-gc.service b/configs/systemd/dist-git-gc.service
new file mode 100644
index 0000000..5203fc5
--- /dev/null
+++ b/configs/systemd/dist-git-gc.service
@@ -0,0 +1,9 @@
+[Unit]
+Description=Git Repositories Clean Up
+
+[Service]
+Type=oneshot
+ExecStart=/usr/bin/dist-git-gc
+
+[Install]
+WantedBy=multi-user.target
diff --git a/configs/systemd/dist-git-gc.timer b/configs/systemd/dist-git-gc.timer
new file mode 100644
index 0000000..fbdfcf9
--- /dev/null
+++ b/configs/systemd/dist-git-gc.timer
@@ -0,0 +1,8 @@
+[Unit]
+Description=Git Repositories Clean Up Timer
+
+[Timer]
+OnCalendar=monthly
+
+[Install]
+WantedBy=timers.target
diff --git a/dist-git.spec b/dist-git.spec
index fc770fd..cb1ceb9 100644
--- a/dist-git.spec
+++ b/dist-git.spec
@@ -165,6 +165,7 @@ ln -s %{_datadir}/dist-git/setup_git_package %{buildroot}%{_bindir}/setup_git_pa
 ln -s %{_datadir}/dist-git/mkbranch %{buildroot}%{_bindir}/mkbranch
 ln -s %{_datadir}/dist-git/mkbranch_branching %{buildroot}%{_bindir}/mkbranch_branching
 ln -s %{_datadir}/dist-git/remove_unused_sources %{buildroot}%{_bindir}/remove_unused_sources
+mv %{buildroot}%{_datadir}/dist-git/dist-git-gc %{buildroot}%{_bindir}/dist-git-gc
 
 # ------------------------------------------------------------------------------
 # SELinux
@@ -226,6 +227,8 @@ fi
 
 %{_unitdir}/dist-git@.service
 %{_unitdir}/dist-git.socket
+%{_unitdir}/dist-git-gc.service
+%{_unitdir}/dist-git-gc.timer
 
 # ------------------------------------------------------------------------------
 # /var/lib/ ...... dynamic persistent files
diff --git a/scripts/dist-git/dist-git-gc b/scripts/dist-git/dist-git-gc
new file mode 100755
index 0000000..4e5547d
--- /dev/null
+++ b/scripts/dist-git/dist-git-gc
@@ -0,0 +1,11 @@
+#!/usr/bin/bash
+# Cleanup unnecessary files
+# The default DEPTH is 2 because the directory path is /var/lib/dist-git/git/rpms/<package>
+
+REPODIR=$(crudini --get dist-git.conf dist-git gitroot_dir)
+DEPTH=$(crudini --get dist-git.conf dist-git git_gc_depth)
+
+find "$REPODIR" -mindepth "$DEPTH" -maxdepth "$DEPTH" -type d |
+  while read -r DIRECTORY; do
+    (cd "$DIRECTORY" && git gc)
+  done