cmd/roachtest: deflake gossip/chaos roachtest

Deflake `gossip/chaos` by adding a missing
`waitForFullReplication`. This test loops, killing a node and then
verifying that the remaining nodes in the cluster stabilize on the same
view of gossip connectivity. Periodically the test was failing because
gossip wasn't stabilizing. The root issue was that the SQL query to
retrieve the gossip connectivity from one node was hanging. And that
query was hanging due to unavailability of a range. Logs show that the
leaseholder for that range was on a down node and that the range only
seemed to contain a single replica. This could happen near the start of
the test if we started killing nodes before full replication was
achieved.

Fixes #38829

Release note: None

pkg/cmd/roachtest/gossip.go

@@ -32,8 +32,10 @@ import (
 
 func registerGossip(r *testRegistry) {
 	runGossipChaos := func(ctx context.Context, t *test, c *cluster) {
+		args := startArgs("--args=--vmodule=*=1")
 		c.Put(ctx, cockroach, "./cockroach", c.All())
-		c.Start(ctx, t, c.All())
+		c.Start(ctx, t, c.All(), args)
+		waitForFullReplication(t, c.Conn(ctx, 1))
 
 		gossipNetwork := func(node int) string {
 			const query = `
@@ -65,6 +67,7 @@ SELECT string_agg(source_id::TEXT || ':' || target_id::TEXT, ',')
 				if i == deadNode {
 					continue
 				}
+				c.l.Printf("%d: checking gossip\n", i)
 				s := gossipNetwork(i)
 				if !initialized {
 					deadNodeStr := fmt.Sprint(deadNode)
@@ -88,7 +91,7 @@ SELECT string_agg(source_id::TEXT || ':' || target_id::TEXT, ',')
 					return false
 				}
 			}
-			fmt.Printf("gossip ok: %s (%0.0fs)\n", expected, timeutil.Since(start).Seconds())
+			c.l.Printf("gossip ok: %s (%0.0fs)\n", expected, timeutil.Since(start).Seconds())
 			return true
 		}
 
@@ -109,7 +112,7 @@ SELECT string_agg(source_id::TEXT || ':' || target_id::TEXT, ',')
 			deadNode = nodes.randNode()[0]
 			c.Stop(ctx, c.Node(deadNode))
 			waitForGossip()
-			c.Start(ctx, t, c.Node(deadNode))
+			c.Start(ctx, t, c.Node(deadNode), args)
 		}
 	}