Suppress locking error messages in monitoring initscripts

We already use --sysinit to disable these messages when using "vgchange -ay" in rc.sysinit stage. However, there's also a "vgchange --monitor y" call later, but this is still too early to expect all locks to have the infrastructure in place (like using cluster locking globally in lvm.conf, but monitoring starts much earlier than cluster infrastructure, so we'd always end with a scary error message at boot).

The user should not see these messages about socket failures and fallback to another type of locking, at least not at this moment. It's misleading - this is actually not a real problem at all, but the messages look like something terrible is happening.

Peter

---

 lib/locking/locking.c                              |    2 +-
 scripts/lvm2_monitoring_init_red_hat.in            |    1 +
 scripts/lvm2_monitoring_systemd_red_hat.service.in |    1 +
 3 files changed, 3 insertions(+), 1 deletions(-)

index 8d709f6..a234ed9 100644
@@ -221,7 +221,7 @@ static void _update_vg_lock_count(const char *resource, uint32_t flags)
  */
 int init_locking(int type, struct cmd_context *cmd, int suppress_messages)
 {
-	if (ignorelockingfailure() && getenv("LVM_SUPPRESS_LOCKING_FAILURE_MESSAGES"))
+	if (getenv("LVM_SUPPRESS_LOCKING_FAILURE_MESSAGES"))
 		suppress_messages = 1;
 
 	if (type < 0)

index 4812c75..0988511 100644
@@ -42,6 +42,7 @@ VGS=${sbindir}/vgs
 LOCK_FILE="/var/lock/subsys/$DAEMON"
 
 WARN=1
+export LVM_SUPPRESS_LOCKING_FAILURE_MESSAGES=1
 
 start()
 {

index 4715416..b5dd9bb 100644
@@ -8,6 +8,7 @@ Conflicts=shutdown.target
 
 [Service]
 Type=oneshot
+Environment=LVM_SUPPRESS_LOCKING_FAILURE_MESSAGES=1
 ExecStart=@sbindir@/lvm vgchange --monitor y
 ExecStop=@sbindir@/lvm vgchange --monitor n
 RemainAfterExit=yes