From 6b22e9c84c00989ea58b1d9df75f2c20eaa2c274 Mon Sep 17 00:00:00 2001

From: Jan Macku <jamacku@redhat.com>

Date: Fri, 17 Mar 2023 14:00:15 +0100

Subject: [PATCH] unit: In cgroupv1, gracefully terminate delegated scopes

 again

Instantly killing delegated scopes is just not viable for our

needs in OCP 4.13 in cgroupv1 mode.  We will accept the possibility

of timeouts instead.

Co-authored-by: Colin Walters <walters@verbum.org>

rhel-only

Resolves: #2179165

---

 src/core/unit.c | 7 ++++++-

 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/src/core/unit.c b/src/core/unit.c

index c319e99d71..0d1a590a3f 100644

--- a/src/core/unit.c

+++ b/src/core/unit.c

@@ -4616,8 +4616,13 @@ int unit_kill_context(

                          * however should not exist in non-delegated units. On the unified hierarchy that's different,

                          * there we get proper events. Hence rely on them. */

+                        /* (RHEL9): we patch out a check for delegation here that exists upstream

+                         * and accept a possible delayed shutdown due to races in favor of

+                         * not just insta-killing the processes.

+                         */

+

                         if (cg_unified_controller(SYSTEMD_CGROUP_CONTROLLER) > 0 ||

-                            (detect_container() == 0 && !unit_cgroup_delegate(u)))

+                            (detect_container() == 0))

                                 wait_for_exit = true;

                         if (send_sighup) {