From cbedbc0ca5d27296b93b89ba2dad3ab16bd50f46 Mon Sep 17 00:00:00 2001
From: Tim Hockin <thockin@google.com>
Date: Sat, 18 Feb 2023 16:26:18 -0800
Subject: [PATCH] Add test for repo size and change default GC

--git-gc=always seems to be the right tradeoff.
---
 cmd/git-sync/main.go | 15 +++++---
 test_e2e.sh          | 87 ++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 96 insertions(+), 6 deletions(-)

diff --git a/cmd/git-sync/main.go b/cmd/git-sync/main.go
index 0982aa9..75d8aa9 100644
--- a/cmd/git-sync/main.go
+++ b/cmd/git-sync/main.go
@@ -136,7 +136,7 @@ var flGitCmd = pflag.String("git", envString("GIT_SYNC_GIT", "git"),
 	"the git command to run (subject to PATH search, mostly for testing)")
 var flGitConfig = pflag.String("git-config", envString("GIT_SYNC_GIT_CONFIG", ""),
 	"additional git config options in 'section.var1:val1,\"section.sub.var2\":\"val2\"' format")
-var flGitGC = pflag.String("git-gc", envString("GIT_SYNC_GIT_GC", "auto"),
+var flGitGC = pflag.String("git-gc", envString("GIT_SYNC_GIT_GC", "always"),
 	"git garbage collection behavior: one of 'auto', 'always', 'aggressive', or 'off'")
 
 var flHTTPBind = pflag.String("http-bind", envString("GIT_SYNC_HTTP_BIND", ""),
@@ -1371,6 +1371,11 @@ func (git *repoSync) cleanup(ctx context.Context, oldHash string) error {
 		}
 	}
 
+	// Expire old refs.
+	if _, err := git.run.Run(ctx, git.root, nil, git.cmd, "reflog", "expire", "--expire-unreachable=all", "--all"); err != nil {
+		cleanupErrs = append(cleanupErrs, err)
+	}
+
 	// Run GC if needed.
 	if git.gc != gcOff {
 		args := []string{"gc"}
@@ -1547,11 +1552,9 @@ func (git *repoSync) SyncRepo(ctx context.Context, refreshCreds func(context.Con
 	// Mark ourselves as "ready".
 	setRepoReady()
 
-	if oldHash != "" {
-		// Clean up the old worktree(s).
-		if err := git.cleanup(ctx, oldHash); err != nil {
-			git.log.Error(err, "git cleanup failed", "oldHash", oldHash)
-		}
+	// Clean up the old worktree(s) and run GC.
+	if err := git.cleanup(ctx, oldHash); err != nil {
+		git.log.Error(err, "git cleanup failed", "oldHash", oldHash)
 	}
 
 	return changed, remote, nil
diff --git a/test_e2e.sh b/test_e2e.sh
index 6fc25a8..095bfc8 100755
--- a/test_e2e.sh
+++ b/test_e2e.sh
@@ -2571,6 +2571,93 @@ function e2e::github_https() {
     assert_file_exists "$ROOT"/link/LICENSE
 }
 
+##############################################
+# Test git-gc default
+##############################################
+function e2e::gc_default() {
+    SHA1=$(git -C "$REPO" rev-parse HEAD)
+    dd if=/dev/urandom of="$REPO"/big1 bs=1024 count=4096 >/dev/null
+    git -C "$REPO" add .
+    git -C "$REPO" commit -qam "$FUNCNAME 1"
+    SHA2=$(git -C "$REPO" rev-parse HEAD)
+    dd if=/dev/urandom of="$REPO"/big2 bs=1024 count=4096 >/dev/null
+    git -C "$REPO" add .
+    git -C "$REPO" commit -qam "$FUNCNAME 2"
+    SHA3=$(git -C "$REPO" rev-parse HEAD)
+
+    GIT_SYNC \
+        --one-time \
+        --repo="file://$REPO" \
+        --root="$ROOT" \
+        --link="link" \
+        --ref="$SHA3" \
+        --depth=0
+    assert_link_exists "$ROOT"/link
+    assert_file_exists "$ROOT"/link/big1
+    assert_file_exists "$ROOT"/link/big2
+    SIZE=$(du -s "$ROOT" | cut -f1)
+    if [ "$SIZE" -lt 14000 ]; then
+        fail "repo is impossibly small: $SIZE"
+    fi
+    if [ "$SIZE" -gt 18000 ]; then
+        fail "repo is too big: $SIZE"
+    fi
+
+    GIT_SYNC \
+        --one-time \
+        --repo="file://$REPO" \
+        --root="$ROOT" \
+        --link="link" \
+        --ref="$SHA3" \
+        --depth=1
+    assert_link_exists "$ROOT"/link
+    assert_file_exists "$ROOT"/link/big1
+    assert_file_exists "$ROOT"/link/big2
+    SIZE=$(du -s "$ROOT" | cut -f1)
+    if [ "$SIZE" -lt 14000 ]; then
+        fail "repo is impossibly small: $SIZE"
+    fi
+    if [ "$SIZE" -gt 18000 ]; then
+        fail "repo is too big: $SIZE"
+    fi
+
+    GIT_SYNC \
+        --one-time \
+        --repo="file://$REPO" \
+        --root="$ROOT" \
+        --link="link" \
+        --ref="$SHA2" \
+        --depth=1
+    assert_link_exists "$ROOT"/link
+    assert_file_exists "$ROOT"/link/big1
+    assert_file_absent "$ROOT"/link/big2
+    SIZE=$(du -s "$ROOT" | cut -f1)
+    if [ "$SIZE" -lt 7000 ]; then
+        fail "repo is impossibly small: $SIZE"
+    fi
+    if [ "$SIZE" -gt 9000 ]; then
+        fail "repo is too big: $SIZE"
+    fi
+
+    GIT_SYNC \
+        --one-time \
+        --repo="file://$REPO" \
+        --root="$ROOT" \
+        --link="link" \
+        --ref="$SHA1" \
+        --depth=1
+    assert_link_exists "$ROOT"/link
+    assert_file_absent "$ROOT"/link/big1
+    assert_file_absent "$ROOT"/link/big2
+    SIZE=$(du -s "$ROOT" | cut -f1)
+    if [ "$SIZE" -lt 100 ]; then
+        fail "repo is impossibly small: $SIZE"
+    fi
+    if [ "$SIZE" -gt 1000 ]; then
+        fail "repo is too big: $SIZE"
+    fi
+}
+
 ##############################################
 # Test git-gc=auto
 ##############################################