github/repmgr
1
0
Fork 0
mirror of https://github.com/EnterpriseDB/repmgr.git synced 2026-03-23 07:06:30 +00:00
Code Issues Packages Projects Releases 1 Wiki Activity

standby follow: don't attempt to delete slot if current upstream is primary

Browse Source 
An attempt will be made to delete an existing replication slot on the
old upstream node (this is important during e.g. a switchover operation
or when attaching a cascaded standby to a new upstream). However if the
standby is currently attached to the follow target node anyway, the
replication slot should never be deleted.
This commit is contained in:
Ian Barwick
2019-12-10 15:48:49 +09:00
parent 0946073406
commit bc93d2996c
1 changed files with 12 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
repmgr-action-standby.c
View File

@@ -2901,8 +2901,8 @@ do_standby_follow_internal(PGconn *primary_conn, PGconn *follow_target_conn, t_n
free_conninfo_params(&local_node_conninfo);
/*
* store the original upstream node id so we can delete the
* replication slot, if exists
* Store the original upstream node id so we can delete the
* replication slot, if it exists.
*/
if (local_node_record.upstream_node_id != UNKNOWN_NODE_ID)
{
@@ -2914,9 +2914,17 @@ do_standby_follow_internal(PGconn *primary_conn, PGconn *follow_target_conn, t_n
}
if (config_file_options.use_replication_slots && runtime_options.host_param_provided == false && original_upstream_node_id != UNKNOWN_NODE_ID)
if (config_file_options.use_replication_slots && runtime_options.host_param_provided == false)
{
remove_old_replication_slot = true;
/*
* Only attempt to delete the old replication slot if the old upstream
* node is known and is different to the follow target node.
*/
if (original_upstream_node_id != UNKNOWN_NODE_ID
&& original_upstream_node_id != follow_target_node_record->node_id)
{
remove_old_replication_slot = true;
}
}
}
@@ -3063,8 +3071,6 @@ do_standby_follow_internal(PGconn *primary_conn, PGconn *follow_target_conn, t_n
* Note that if this function is called by do_standby_switchover(), the
* "repmgr node rejoin" command executed on the demotion candidate may already
* have removed the slot, so there may be nothing to do.
*
* XXX check if former upstream is current primary?
*/
if (remove_old_replication_slot == true)