diff --git a/mercurial/setdiscovery.py b/mercurial/setdiscovery.py
--- a/mercurial/setdiscovery.py
+++ b/mercurial/setdiscovery.py
@@ -105,10 +105,7 @@ def _takequicksample(dag, nodes, size):
     return sample
 
 def _takefullsample(dag, nodes, size):
-    always = dag.headsetofconnecteds(nodes)
-    if size <= len(always):
-        return always
-    sample = always
+    sample = always = dag.headsetofconnecteds(nodes)
     # update from heads
     _updatesample(dag, nodes, sample, always)
     # update from roots
diff --git a/tests/test-setdiscovery.t b/tests/test-setdiscovery.t
--- a/tests/test-setdiscovery.t
+++ b/tests/test-setdiscovery.t
@@ -317,17 +317,14 @@ One with >200 heads, which used to use u
   query 3; still undecided: 1140, sample size is: 200
   sampling from both directions
   searching: 4 queries
-  query 4; still undecided: 940, sample size is: 200
+  query 4; still undecided: 592, sample size is: 200
   sampling from both directions
   searching: 5 queries
-  query 5; still undecided: 740, sample size is: 200
+  query 5; still undecided: 292, sample size is: 200
   sampling from both directions
   searching: 6 queries
-  query 6; still undecided: 540, sample size is: 200
-  sampling from both directions
-  searching: 7 queries
-  query 7; still undecided: 46, sample size is: 46
-  7 total queries
+  query 6; still undecided: 51, sample size is: 51
+  6 total queries
   common heads: 3ee37d65064a
 
 Test actual protocol when pulling one new head in addition to common heads