From: Amitay Isaacs Date: Mon, 11 Apr 2016 04:01:42 +0000 (+1000) Subject: ctdb-tests: Fix flakey test complex/18_ctdb_reloadips.sh X-Git-Tag: talloc-2.1.7~205 X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=fdaa2310ffc1f74da48c5354097400955030f509;p=thirdparty%2Fsamba.git ctdb-tests: Fix flakey test complex/18_ctdb_reloadips.sh This test sets TakeoverTimeout=90 to avoid banning during takeover. However, the setting is done on the test node instead of the recovery master node. During "ctdb reloadips", the recovery master will used the default value of TakeoverTimeout. Signed-off-by: Amitay Isaacs Reviewed-by: Martin Schwenke --- diff --git a/ctdb/tests/complex/18_ctdb_reloadips.sh b/ctdb/tests/complex/18_ctdb_reloadips.sh index 9a408353901..c61dcdc7c6d 100755 --- a/ctdb/tests/complex/18_ctdb_reloadips.sh +++ b/ctdb/tests/complex/18_ctdb_reloadips.sh @@ -93,9 +93,10 @@ iface=$(echo "$ctdb_ip_info" | awk -v pnn=$test_node '$2 == pnn { print $3 ; exi #################### +# This needs to be set only on the recmaster. All nodes should do the trick. new_takeover_timeout=90 echo "Setting TakeoverTimeout=${new_takeover_timeout} to avoid potential bans" -try_command_on_node $test_node "$CTDB setvar TakeoverTimeout ${new_takeover_timeout}" +try_command_on_node all "$CTDB setvar TakeoverTimeout ${new_takeover_timeout}" ####################