Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.1.25, 10.2.12, 10.2.13
-
QA environment, 3 Node MariaDB Galera Cluster - 10.2.12 + galera-25.3.22, Amazon AMI 2017.9
1.3TB worth of data
QA Environment, 3 Node MariaDB Galera Cluster - 10.1.25 + galera-25.3.20, Amazon AMI 2017.3
1.3TB worth of data
Description
I'm running the following procedure on one node in a 3 node galera cluster:
mysql -u root -p
|
use {foobar};
|
SET SESSION wsrep_OSU_method='RSU';
|
SET GLOBAL slave_type_conversions='ALL_LOSSY';
|
ALTER TABLE {foo} DROP FOREIGN KEY fkb1;
|
ALTER TABLE {foo} MODIFY COLUMN {column_name} BIGINT UNSIGNED NOT NULL AUTO_INCREMENT, MODIFY COLUMN {second_column} BIGINT UNSIGNED;
|
ALTER TABLE {foo} ADD CONSTRAINT `fkb1` FOREIGN KEY {second column} REFERENCES `{foobar}` (`{foo}`) ON DELETE NO ACTION ON UPDATE NO ACTION;
|
SET SESSION wsrep_OSU_method='TOI';
|
The change works as expected, however, when trying to sync the node back to the cluster, the following occurs:
2018-02-13 22:52:03 139478253385472 [Note] Start binlog_dump to slave_server(0), pos(nodename-bin.000011, 4)
|
2018-02-13 22:52:07 139478327580416 [ERROR] Slave SQL: Column 0 of table '{foobar}.{foo}' cannot be converted from type 'int' to type 'bigint(20) unsigned', Internal MariaDB error code: 1677
|
2018-02-13 22:52:07 139478327580416 [Warning] WSREP: RBR event 3 Update_rows_v1 apply warning: 3, 15861831987
|
2018-02-13 22:52:07 139478327580416 [Warning] WSREP: Failed to apply app buffer: seqno: 15861831987, status: 1
|
at galera/src/trx_handle.cpp:apply():351
|
Retrying 2th time
|
2018-02-13 22:52:07 139478327580416 [ERROR] Slave SQL: Column 0 of table '{foobar}.{foo}' cannot be converted from type 'int' to type 'bigint(20) unsigned', Internal MariaDB error code: 1677
|
2018-02-13 22:52:07 139478327580416 [Warning] WSREP: RBR event 3 Update_rows_v1 apply warning: 3, 15861831987
|
2018-02-13 22:52:07 139478327580416 [Warning] WSREP: Failed to apply app buffer: seqno: 15861831987, status: 1
|
at galera/src/trx_handle.cpp:apply():351
|
Retrying 3th time
|
2018-02-13 22:52:07 139478327580416 [ERROR] Slave SQL: Column 0 of table '{foobar}.{foo}' cannot be converted from type 'int' to type 'bigint(20) unsigned', Internal MariaDB error code: 1677
|
2018-02-13 22:52:07 139478327580416 [Warning] WSREP: RBR event 3 Update_rows_v1 apply warning: 3, 15861831987
|
2018-02-13 22:52:07 139478327580416 [Warning] WSREP: Failed to apply app buffer: seqno: 15861831987, status: 1
|
at galera/src/trx_handle.cpp:apply():351
|
Retrying 4th time
|
2018-02-13 22:52:07 139478327580416 [ERROR] Slave SQL: Column 0 of table '{foobar}.{foo}' cannot be converted from type 'int' to type 'bigint(20) unsigned', Internal MariaDB error code: 1677
|
2018-02-13 22:52:07 139478327580416 [Warning] WSREP: RBR event 3 Update_rows_v1 apply warning: 3, 15861831987
|
2018-02-13 22:52:07 139478327580416 [ERROR] WSREP: Failed to apply trx: source: 517b5eed-10d4-11e8-9200-9ed4847004dd version: 3 local: 0 state: APPLYING flags: 1 conn_id: 2438 trx_id: 25630496491 seqnos (l: 13, g: 15861831987, s: 15861831986, d: 15861831986, ts: 1168584749936241)
|
2018-02-13 22:52:07 139478327580416 [ERROR] WSREP: Failed to apply trx 15861831987 4 times
|
2018-02-13 22:52:07 139478327580416 [ERROR] WSREP: Node consistency compromised, aborting...
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: Closing send monitor...
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: Closed send monitor.
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: gcomm: terminating thread
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: gcomm: joining thread
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: gcomm: closing backend
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: view(view_id(NON_PRIM,1f26c4fc,3) memb {
|
1f26c4fc,0
|
} joined {
|
} left {
|
} partitioned {
|
517b5eed,0
|
70b64642,0
|
})
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: view((empty))
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: gcomm: closed
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: New COMPONENT: primary = no, bootstrap = no, my_idx = 0, memb_num = 1
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: Flow-control interval: [253, 256]
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: Trying to continue unpaused monitor
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: Received NON-PRIMARY.
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: Shifting SYNCED -> OPEN (TO: 15861831987)
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: Received self-leave message.
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: Flow-control interval: [253, 256]
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: Trying to continue unpaused monitor
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: Received SELF-LEAVE. Closing connection.
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: Shifting OPEN -> CLOSED (TO: 15861831987)
|
2018-02-13 22:52:07 139622020704000 [Note] WSREP: RECV thread exiting 0: Success
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: recv_thread() joined.
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: Closing replication queue.
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: Closing slave action queue.
|
2018-02-13 22:52:07 139478327580416 [Note] WSREP: /usr/sbin/mysqld: Terminated.
|
This is after having set slave_type_conversions=ALL_LOSSY; on all Galera cluster nodes.
Reference to MDEV-9490 where this issue was apparently fixed - we're still seeing these problems in MariaDB 10.2.12 and 10.1.25
Attachments
Issue Links
- relates to
-
-
- Closed
-
-
MDEV-15360 add slave_type_conversions value
-
- Open
-