Details
-
New Feature
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
None
-
None
-
any version
Description
Maxscale requires GTID repl for failover but servers show gtid by default. I would like a better display of information on this since if a slave is rebuilt and misconfigured with basic binlog_pos it at a glance looks like maxscale is heathy but is in fact no longer HA. This is disastrous if a master goes and stays down to get a replica promoted.
For usability a display like below is ideal to see what the current service status really is
[ACME] root@maxscale-1: etc # maxctrl list servers
|
┌───────────┬──────────────┬──────┬─────────────┬─────────────────┬────────┬─────────────────┐
|
│ Server │ Address │ Port │ Connections │ State │ GTID │ Monitor │
|
├───────────┼──────────────┼──────┼─────────────┼─────────────────┼────────┼─────────────────┤
|
│ fooshop-1 │ 131.21.1.185 │ 3306 │ 0 │ Master, Running │ 0-1-90 │ MariaDB-Monitor │
|
├───────────┼──────────────┼──────┼─────────────┼─────────────────┼────────┼─────────────────┤
|
│ fooshop-2 │ 131.21.1.248 │ 3306 │ 0 │ Slave, Running │ NOT USING GTID REPL │ MariaDB-Monitor │
|
├───────────┼──────────────┼──────┼─────────────┼─────────────────┼────────┼─────────────────┤
|
│ fooshop-3 │ 131.21.1.89 │ 3306 │ 0 │ Slave, Running │ NOT USING GTID REPL │ MariaDB-Monitor │
|
└───────────┴──────────────┴──────┴─────────────┴─────────────────┴────────┴─────────────────┘
|
The slaves here are clearly configured but ineligible for failover
Developer Description
The GTID alone doesn't show all of the information related to replication. Extending the --fields flag from MXS-5220 to include the SQL and IO thread states and other bits from the slave_connections field would be useful:
"slave_connections": [
|
{
|
"connection_name": "",
|
"gtid_io_pos": "",
|
"last_io_error": "",
|
"last_sql_error": "",
|
"master_host": "127.0.0.1",
|
"master_port": 3000,
|
"master_server_id": 3000,
|
"master_server_name": "server1",
|
"seconds_behind_master": 0,
|
"slave_io_running": "Yes",
|
"slave_sql_running": "Yes",
|
"using_gtid": "No"
|
}
|
],
|
Attachments
Issue Links
- relates to
-
-
- Closed
-
The GTID is the GTID position, not how the GTIDs are being replicated. Not displaying it for the cases where file+position replication is used would break it for all users who use it. I think this would be better solved by adding the information in maxctrl show services and show monitors (have you checked these already?) where the other fine details of replication are located.