From 5fcd385572ccf2d4a3c3030723617eaec8ae1516 Mon Sep 17 00:00:00 2001 From: Mark Reynolds Date: Tue, 21 Apr 2015 12:23:47 -0400 Subject: [PATCH 314/319] Ticket 48154 - abort cleanAllRUV tasks should not certify-all by default Bug Description: The current default for an abort task is to certify that all the replicas receive and process the abort task. The problem is that if we are trying to abort a cleanallruv task because it is "hanging" due to some replicas not being reachable, the abort task will also "hang". Fix Description: The default for certify-all should be "no". https://fedorahosted.org/389/ticket/48154 Reviewed by: nhosoi(Thanks!) (cherry picked from commit 465e1059493a8dacf3caf7776824e8149e878b93) (cherry picked from commit ba488271c51a0bd566a4f728709ff959d64a0ed3) --- ldap/servers/plugins/replication/repl5_replica_config.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/ldap/servers/plugins/replication/repl5_replica_config.c b/ldap/servers/plugins/replication/repl5_replica_config.c index 94ab71b..daefc03 100644 --- a/ldap/servers/plugins/replication/repl5_replica_config.c +++ b/ldap/servers/plugins/replication/repl5_replica_config.c @@ -2596,7 +2596,13 @@ replica_cleanall_ruv_abort(Slapi_PBlock *pb, Slapi_Entry *e, Slapi_Entry *eAfter goto out; } } else { - certify_all = "yes"; + /* + * The default should be not to certify all the replicas, because + * we might be trying to abort a clean task that is "hanging" due + * to unreachable replicas. If the default is "yes" then the abort + * task will run into the same issue. + */ + certify_all = "no"; } /* * Create payload -- 1.9.3