Fix timing issues in flaky scaledown tests (#189)

Terminate a clusterclient directly instead of first closing its client
socket. This avoids that hiredis-cluster might succeed to reconnect to
the server before the server has shutdown.

Close the simulated-redis's listener socket early improves stability.

Add check for alternative output depending on timing in following testcases:
cluster-scale-down-test.sh
dbsize-to-all-nodes-during-scaledown-test.sh
dbsize-to-all-nodes-during-scaledown-test-async.sh

tests/scripts/cluster-scale-down-test.sh

@@ -1,4 +1,4 @@
-#!/bin/sh
+#!/bin/bash
 #
 # To verify the clients behaviour in a cluster scaledown scenario.
 # The testcase will send commands, all targeting hash slot 12182, while removing
@@ -64,7 +64,6 @@ EXPECT CONNECT
 EXPECT ["GET", "{foo}1"]
 SEND "bar1"
 # Forced close. The next command "GET {foo}2" will fail.
-CLOSE
 EOF
 server2=$!
 
@@ -97,12 +96,20 @@ if [ $clientexit -ne 0 ]; then
     exit $clientexit
 fi
 
-# Check the output from clusterclient
-expected="bar1
+# Check the output from clusterclient, which depends on timing.
+# Client sends the command 'GET {foo}2' just after nodeid2 closes its socket.
+expected1="bar1
 error: Server closed the connection
 bar3"
 
-echo "$expected" | diff -u - "$testname.out" || exit 99
+# Client sends the command 'GET {foo}2' just before nodeid2 closes its socket.
+expected2="bar1
+error: Connection reset by peer
+bar3"
+
+diff -u "$testname.out" <(echo "$expected1") || \
+    diff -u "$testname.out" <(echo "$expected2") || \
+    exit 99
 
 # Clean up
 rm "$testname.out"

tests/scripts/dbsize-to-all-nodes-during-scaledown-test-async.sh

@@ -31,7 +31,7 @@ EXPECT ["DBSIZE"]
 SEND 10
 EXPECT ["DBSIZE"]
 SEND 11
-# The second command to node2 fails which triggers a slotmap update.
+# The second command to node #2 fails which triggers a slotmap update.
 EXPECT ["CLUSTER", "SLOTS"]
 SEND [[0, 16383, ["127.0.0.1", 7401, "nodeid7401"]]]
 EXPECT ["DBSIZE"]
@@ -45,7 +45,7 @@ timeout 5s ./simulated-redis.pl -p 7402 -d --sigcont $syncpid2 <<'EOF' &
 EXPECT CONNECT
 EXPECT ["DBSIZE"]
 SEND 20
-CLOSE
+# Forced close. The second command to this node should trigger a slotmap update.
 EOF
 server2=$!
 
@@ -81,14 +81,26 @@ if [ $clientexit -ne 0 ]; then
     exit $clientexit
 fi
 
-# Check the output from clusterclient
-expected="10
+# Check the output from clusterclient, which depends on timing.
+# Client sends the second 'DBSIZE' to node #2 just after node #2 closes its socket.
+expected1="10
 20
 error: Server closed the connection
 11
 12"
 
-echo "$expected" | diff -u - "$testname.out" || exit 99
+# Client sends the second 'DBSIZE' to node #2 just before node #2 closes its socket.
+expected2="10
+20
+error: Connection reset by peer
+11
+12"
+
+# The reply "11" from node #1 can come before or after the socket error from node #2.
+# Therefore, we sort before comparing.
+diff -u <(echo "$expected1" | sort) <(sort "$testname.out") || \
+    diff -u <(echo "$expected2" | sort) <(sort "$testname.out") || \
+    exit 99
 
 # Clean up
 rm "$testname.out"

tests/scripts/dbsize-to-all-nodes-during-scaledown-test.sh

@@ -33,7 +33,7 @@ EXPECT ["DBSIZE"]
 SEND 11
 EXPECT ["DBSIZE"]
 SEND 12
-# The second command to node2 fails which triggers a slotmap update pipelined
+# The second command to node #2 fails which triggers a slotmap update pipelined
 # onto the 3rd DBSIZE to this node.
 EXPECT ["CLUSTER", "SLOTS"]
 SEND [[0, 16383, ["127.0.0.1", 7401, "nodeid7401"]]]
@@ -46,7 +46,7 @@ timeout 5s ./simulated-redis.pl -p 7402 -d --sigcont $syncpid2 <<'EOF' &
 EXPECT CONNECT
 EXPECT ["DBSIZE"]
 SEND 20
-CLOSE
+# Forced close. The second command to this node should trigger a slotmap update.
 EOF
 server2=$!
 
@@ -82,14 +82,24 @@ if [ $clientexit -ne 0 ]; then
     exit $clientexit
 fi
 
-# Check the output from clusterclient
-expected="10
+# Check the output from clusterclient, which depends on timing.
+# Client sends the second 'DBSIZE' to node #2 just after node #2 closes its socket.
+expected1="10
 20
 11
 error: Server closed the connection
 12"
 
-echo "$expected" | diff -u - "$testname.out" || exit 99
+# Client sends the second 'DBSIZE' to node #2 just before node #2 closes its socket.
+expected2="10
+20
+11
+error: Connection reset by peer
+12"
+
+diff -u "$testname.out" <(echo "$expected1") || \
+    diff -u "$testname.out" <(echo "$expected2") || \
+    exit 99
 
 # Clean up
 rm "$testname.out"

tests/scripts/simulated-redis.pl

@@ -166,6 +166,7 @@ END
         unexpected($port, "event: $_");
     }
 }
+close $listener;
 print "(port $port) Done.\n" if $debug;
 exit;