File pacemaker-libpe_status-detect-when-nodes-should-suicide.patch of Package pacemaker.14737

commit 4f574e10e3ce4838caf086180a8cb8d9b637b722
Author: Ken Gaillot <kgaillot@redhat.com>
Date:   Thu Oct 5 15:23:22 2017 -0500

    Low: libpe_status: properly detect when nodes should suicide
    
    64fdcf44 left no-quorum-policy as suicide when the cluster had not previously
    had quorum, but did now -- which triggered fencing code that should have
    been checking whether quorum was lost but wasn't. That is now fixed.

diff --git a/lib/pengine/unpack.c b/lib/pengine/unpack.c
index 141ab5e49..8cb2f1307 100644
--- a/lib/pengine/unpack.c
+++ b/lib/pengine/unpack.c
@@ -1210,7 +1210,9 @@ unpack_status(xmlNode * status, pe_working_set_t * data_set)
             crm_trace("determining node state");
             determine_online_status(state, this_node, data_set);
 
-            if (this_node->details->online && data_set->no_quorum_policy == no_quorum_suicide) {
+            if (is_not_set(data_set->flags, pe_flag_have_quorum)
+                && this_node->details->online
+                && (data_set->no_quorum_policy == no_quorum_suicide)) {
                 /* Everything else should flow from this automatically
                  * At least until the PE becomes able to migrate off healthy resources
                  */