Merge pull request #468: maintenance: delete stale lock files, fix lo…

…ose-objects task

The maintenance.lock file exists to prevent concurrent maintenance
processes from writing to a repository at the same time. However, it has
the downside of causing maintenance to start failing without recovery if
there is any reason why the maintenance command failed without cleaning
up the lock-file.

This change makes it such that maintenance will delete a lock file that
was modified over 6 hours ago. This will auto-heal repositories that are
stuck with failed maintenance (and maybe it will fail again, but we will
get a message other than the lock file exists).

The intention here is to get users out of a bad state without weakening the maintenance lock _too_ much. I'm open to other thoughts, including expanding the timeframe from 6 to 24 hours.

---

The `fixup!` commit fixes the use of the `gvfs.sharedcache` config value around the loose-objects task. Specifically, the loose-objects step might fail because it is calling `git pack-objects {garbage}/pack/loose` which fails.

builtin/gc.c

@@ -1274,6 +1274,8 @@ static int maintenance_run_tasks(struct maintenance_run_opts *opts)
 	char *lock_path = xstrfmt("%s/maintenance", r->objects->odb->path);
 
 	if (hold_lock_file_for_update(&lk, lock_path, LOCK_NO_DEREF) < 0) {
+		struct stat st;
+		struct strbuf lock_dot_lock = STRBUF_INIT;
 		/*
 		 * Another maintenance command is running.
 		 *
@@ -1284,6 +1286,25 @@ static int maintenance_run_tasks(struct maintenance_run_opts *opts)
 		if (!opts->auto_flag && !opts->quiet)
 			warning(_("lock file '%s' exists, skipping maintenance"),
 				lock_path);
+
+		/*
+		 * Check timestamp on .lock file to see if we should
+		 * delete it to recover from a fail state.
+		 */
+		strbuf_addstr(&lock_dot_lock, lock_path);
+		strbuf_addstr(&lock_dot_lock, ".lock");
+		if (lstat(lock_dot_lock.buf, &st))
+			warning_errno(_("unable to stat '%s'"), lock_dot_lock.buf);
+		else {
+			if (st.st_mtime < time(NULL) - (6 * 60 * 60)) {
+				if (unlink(lock_dot_lock.buf))
+					warning_errno(_("unable to delete stale lock file"));
+				else
+					warning(_("deleted stale lock file"));
+			}
+		}
+
+		strbuf_release(&lock_dot_lock);
 		free(lock_path);
 		return 0;
 	}
@@ -1412,6 +1433,7 @@ static int maintenance_run(int argc, const char **argv, const char *prefix)
 {
 	int i;
 	struct maintenance_run_opts opts;
+	const char *tmp_obj_dir = NULL;
 	struct option builtin_maintenance_run_options[] = {
 		OPT_BOOL(0, "auto", &opts.auto_flag,
 			 N_("run tasks based on the state of the repository")),
@@ -1451,9 +1473,11 @@ static int maintenance_run(int argc, const char **argv, const char *prefix)
 	 * the gvfs.sharedcache config option to redirect the
 	 * maintenance to that location.
 	 */
-	if (!git_config_get_value("gvfs.sharedcache", &object_dir) &&
-	    object_dir)
+	if (!git_config_get_value("gvfs.sharedcache", &tmp_obj_dir) &&
+	    tmp_obj_dir) {
+		object_dir = xstrdup(tmp_obj_dir);
 		setenv(DB_ENVIRONMENT, object_dir, 1);
+	}
 
 	return maintenance_run_tasks(&opts);
 }

t/t7900-maintenance.sh

@@ -52,6 +52,23 @@ test_expect_success 'run [--auto|--quiet]' '
 	test_subcommand git gc --no-quiet <run-no-quiet.txt
 '
 
+test_expect_success 'lock file behavior' '
+	test_when_finished git config --unset maintenance.commit-graph.schedule &&
+	git config maintenance.commit-graph.schedule hourly &&
+
+	touch .git/objects/maintenance.lock &&
+	git maintenance run --schedule=hourly --no-quiet 2>err &&
+	grep "lock file .* exists, skipping maintenance" err &&
+
+	test-tool chmtime =-22000 .git/objects/maintenance.lock &&
+	git maintenance run --schedule=hourly --no-quiet 2>err &&
+	grep "deleted stale lock file" err &&
+	test_path_is_missing .git/objects/maintenance.lock &&
+
+	git maintenance run --schedule=hourly 2>err &&
+	test_must_be_empty err
+'
+
 test_expect_success 'maintenance.auto config option' '
 	GIT_TRACE2_EVENT="$(pwd)/default" git commit --quiet --allow-empty -m 1 &&
 	test_subcommand git maintenance run --auto --quiet <default &&