From f20aa782d642bbdb1119667d25928739c9a83c89 Mon Sep 17 00:00:00 2001 From: Ludwig Krispenz Date: Jan 11 2018 15:52:23 +0000 Subject: Ticket 49446 - cleanallruv should ignore cleaned replica Id in processing changelog if in force mode Bug: If the startcsn is calculated based on a cleaned rid, it could be missing from the changelog. Fix: In force mode we do not care that the topology gets in sync for the cleaned RID, so we can ignore it in an earlier stage, instead of setting it to precleane only. Reviewed by: Thierry, thanks --- diff --git a/ldap/servers/plugins/replication/repl5_replica_config.c b/ldap/servers/plugins/replication/repl5_replica_config.c index e025f34..005528a 100644 --- a/ldap/servers/plugins/replication/repl5_replica_config.c +++ b/ldap/servers/plugins/replication/repl5_replica_config.c @@ -1688,9 +1688,15 @@ replica_cleanallruv_thread(void *arg) } /* * Presetting the rid prevents duplicate thread creation, but allows the db and changelog to still - * process updates from the rid. set_cleaned_rid() blocks updates, so we don't want to do that... yet. + * process updates from the rid. + * set_cleaned_rid() blocks updates, so we don't want to do that... yet unless we are in force mode. + * If we are forcing a clean independent of state of other servers for this RID we can set_cleaned_rid() */ - preset_cleaned_rid(data->rid); + if (data->force) { + set_cleaned_rid(data->rid); + } else { + preset_cleaned_rid(data->rid); + } rid_text = slapi_ch_smprintf("%d", data->rid); csn_as_string(data->maxcsn, PR_FALSE, csnstr); /*