pxc-scheduler-handler options¶
The pxc_scheduler_handler script lists the available options in the pxc_scheduler_handler configuration file. The following options are described in more detail:
-–add-query-rule¶
This option creates query rules for a synced MySQL user and applies only to the
singlewrite
mode.
--add-query-rule requires¶
Either the –syncusers or –sync-multi-cluster-users options.
--add-query-rule example¶
$ percona-scheduler-admin --config-file=config.toml --syncusers
Expected output
--add-query-rule
Syncing user accounts from PXC to ProxySQL
Note : 'admin' is in proxysql admin user list, this user cannot be added to ProxySQL
-- (For more info, see https://github.com/sysown/proxysql/issues/709)
Adding user to ProxySQL: test_query_rule
Added query rule for user: test_query_rule
Adding user to ProxySQL: test_query_rule
Added query rule for user: test_query_rule
Synced PXC users to the ProxySQL database!
–-adduser¶
This option adds the cluster application user account to the ProxySQL database.
–-adduser example¶
$ percona-scheduler-admin --config-file=config.toml --adduser
Expected output
Adding PXC application user to the ProxySQL database
Enter the PXC application user name: cluster_one
Enter the PXC application user password:
The application user 'cluster_one' does not exist in PXC. Would you like to proceed [y/n] ? y
Please create the user cluster_one in PXC to access the application through ProxySQL
Added PXC application user to the ProxySQL database!
–-auto-assign-weights¶
ProxySQL uses weights for defining the failover procedure in singlewrite
mode
and handling load balancing loadbal
mode.
For the failover procedure, this option with the
--update-cluster option assigns weights to the PXC nodes when the cluster is
in singlewrite
mode.
As a best practice, ensure that the writer node election operation returns the
same result each time. For example, assign the value of 1000
to node one,
999
to node two, and 998
to the node three. This method sets a clear
priority for the election.
For load balancing you want to reduce the reads on the writer node and, also, split the reads across all the reader nodes equally.
For example, in a three-node cluster, assign a 900
to the writer node and
1000
and 1000
to the reader nodes.
This option does these operations automatically without any manual intervention.
Review do not combine certain options.
The following example is a default configuration when the
pxc_scheduler_handler
sets up proxysql.
Expected output
Cluster node info
+---------------+-------+---------------+------+--------+--------+
| hostgroup | hg_id | hostname | port | status | weight |
+---------------+-------+---------------+------+--------+--------+
| writer | 100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
+---------------+-------+---------------+------+--------+--------+
Cluster membership updated in the ProxySQL database!
--auto-assign-weights example¶
$ percona-scheduler-admin --config-file=config.toml --update-cluster --auto-assign-weights
Expected output
No new nodes detected.
Cluster node info
+---------------+-------+---------------+------+--------+--------+
| hostgroup | hg_id | hostname | port | status | weight |
+---------------+-------+---------------+------+--------+--------+
| writer | 100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.32 | 3306 | ONLINE | 900 |
| reader | 101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.34 | 3306 | ONLINE | 998 |
| writer-config | 8100 | 192.168.56.33 | 3306 | ONLINE | 999 |
| writer-config | 8100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.32 | 3306 | ONLINE | 900 |
| reader-config | 8101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
+---------------+-------+---------------+------+--------+--------+
Cluster membership updated in the ProxySQL database!
--config-file¶
This option reads the login credentials from a configuration file. Command-line options override configuration file values. For more information, see pxc_scheduler_handler configuration.
--debug¶
This option enables additional debug logging.
–-disable / -d¶
This option removes Percona XtraDB Cluster nodes from ProxySQL and stops the ProxySQL monitoring daemon.
--disable example¶
$ percona-scheduler-admin --config-file=config.toml --disable
Expected output
Removing cluster application users from the ProxySQL database.
Removing cluster nodes from the ProxySQL database.
Removing query rules from the ProxySQL database if any.
--disable_updates¶
This option can be used with any command to disable updating the
pxc_scheduler_handler checksum for the mysql_query_rules
,
mysql_server
and mysql_users
tables. The option avoids updates in
those tables for one node from being propagated to the cluster.
Specifying this option sets the following values to false
:
- admin-checksum_mysql_query_rules
- admin-checksum_mysql_servers
- admin_checksum_mysql_users
–-enable / -e¶
This option creates the entries for the Galera hostgroups and adds the
Percona XtraDB Cluster nodes into ProxySQL’s mysql_servers
table.
The option adds two users into the Percona XtraDB Cluster with the USAGE
privilege. The users have the following tasks:
-
Monitor the cluster nodes through ProxySQL.
-
Connect to the PXC Cluster node through the ProxySQL console.
You must have super
user credentials from Percona XtraDB Cluster to set up the default users.
--enable example¶
$ percona-scheduler-admin --config-file=config.toml --enable
Expected output
Configuring using mode: singlewrite
The ClusterApp User or Password was unspecified and will not be configured.
This script will assist with configuring ProxySQL for use with
Percona XtraDB Cluster (PXC) (currently only PXC in combination with
ProxySQL is supported)
The ProxySQL read/write configuration mode is singlewrite
Configuring the ProxySQL monitoring user.
ProxySQL monitor username as per command line/config-file is monitor
Monitoring user 'monitor'@'127.%' has been setup in the ProxySQL database.
Adding the Percona XtraDB Cluster nodes to ProxySQL
Using the scheduler binary located at /home/venki/work/proxysql/proxysql-admin-tool/pxc_scheduler_handler
Waiting for scheduler script to process new nodes...
Proxysql status (mysql_servers rows) for this configuration
+---------------+-------+---------------+------+--------+--------+----------+---------+-----------+
| hostgroup | hg_id | hostname | port | status | weight | max_conn | use_ssl | gtid_port |
+---------------+-------+---------------+------+--------+--------+----------+---------+-----------+
| writer | 100 | 192.168.56.32 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader | 101 | 192.168.56.32 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader | 101 | 192.168.56.33 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader | 101 | 192.168.56.34 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| writer-config | 8100 | 192.168.56.32 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| writer-config | 8100 | 192.168.56.33 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| writer-config | 8100 | 192.168.56.34 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader-config | 8101 | 192.168.56.32 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader-config | 8101 | 192.168.56.33 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader-config | 8101 | 192.168.56.34 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
+---------------+-------+---------------+------+--------+--------+----------+---------+-----------+
ProxySQL configuration completed!
ProxySQL has been successfully configured to use with Percona XtraDB Cluster
Verify -enable example¶
You can verify the successful configuration with the following command:
mysql> select * from runtime_mysql_servers;
Expected output
hostgroup_id | hostname | port | gtid_port | status | weight | compression | max_connections | max_replication_lag | use_ssl | max_latency_ms | comment |
100 | 192.168.56.32 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 | |
8101 | 192.168.56.33 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 | |
8101 | 192.168.56.34 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 | |
8101 | 192.168.56.32 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 | |
8100 | 192.168.56.33 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 | |
8100 | 192.168.56.34 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 | |
8100 | 192.168.56.32 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 | |
101 | 192.168.56.33 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 | |
101 | 192.168.56.34 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 | |
101 | 192.168.56.32 | 3306 | 0 | ONLINE | 1000 | 0 | 1000 | 0 | 0 | 0 |
Verify the scheduler settings with the following command:
mysql> select * from scheduler\G
Expected output
*************************** 1. row ***************************
id: 6
active: 1
interval_ms: 5000
filename: <path/to/pxc_scheduler>/pxc_scheduler_handler
arg1: --configfile=config.toml
arg2: --configpath=<path/to/config/dir>
arg3: NULL
arg4: NULL
arg5: NULL
comment: { hgW:100, hgR:101 }
1 row in set (0.00 sec)
–-force¶
This option must be combined with either –enable / -e or
–update-cluster. This option skips any mysql_servers
table,
mysql_users
table, and mysql_galera_hostgroups
table configuration
checks. Certain checks issue warnings, instead of errors, which allows the option to continue processing.
--help¶
This option displays the help text.
–-is-enabled¶
This option checks if pxc_scheduler_handler
configured the hostgroups in ProxySQL.
Returns a zero (0) if an entry corresponds to the writer hostgroup and is set to active in ProxySQL.
Returns a one (1) if an entry does not correspond to the writer hostgroup.
--is-enabled example¶
$ percona-scheduler-admin --config-file=config.toml --is-enabled
Expected output
The current configuration has been enabled and is active
--is-enabled verification example¶
Verify if pxc_scheduler_handler
configured the hostgroups in ProxySQL.
$ echo $?
Expected output
0
--is-enabled example with cluster disabled¶
Disabling the cluster configuration and then calling the option creates an error.
Disabling the cluster:
$ percona-scheduler-admin --config-file=config.toml --disable
Expected output
Removing cluster application users from the ProxySQL database.
Removing cluster nodes from the ProxySQL database.
Removing query rules from the ProxySQL database if any.
ProxySQL configuration removed!
Check if the cluster is enabled:
$ percona-scheduler-admin --config-file=config.toml --is-enabled
Expected output
ERROR (line:2450) : The current configuration has not been enabled
–-remove-all-servers¶
When used with --update-cluster this option removes all servers belonging to the current cluster before running the --update-cluster option.
--remove-all-server example¶
$ percona-scheduler-admin --config-file=config.toml
--remove-all-servers --update-cluster
–-server¶
Selects a server by the IP address and port. This option can be combined with --syncusers or –sync-multi-cluster-users to sync a single non-cluster server. The option does not require that the server belong to a Percona XtraDB Cluster.
-server example¶
$ percona-scheduler-admin --config-file=config.toml --server=192.168.56.32:3306
–-status¶
This option displays information about all Galera hostgroups and their servers supported by this ProxySQL instance.
--status example¶
$ percona-scheduler-admin --config-file=config.toml --status
Expected output
mysql_servers rows for this configuration
+---------------+-------+---------------+------+--------+--------+----------+---------+-----------+
| hostgroup | hg_id | hostname | port | status | weight | max_conn | use_ssl | gtid_port |
+---------------+-------+---------------+------+--------+--------+----------+---------+-----------+
| writer | 100 | 192.168.56.32 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader | 101 | 192.168.56.32 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader | 101 | 192.168.56.33 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader | 101 | 192.168.56.34 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| writer-config | 8100 | 192.168.56.32 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| writer-config | 8100 | 192.168.56.33 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| writer-config | 8100 | 192.168.56.34 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader-config | 8101 | 192.168.56.32 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader-config | 8101 | 192.168.56.33 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
| reader-config | 8101 | 192.168.56.34 | 3306 | ONLINE | 1000 | 1000 | 0 | 0 |
+---------------+-------+---------------+------+--------+--------+----------+---------+-----------+
–-sync-multi-cluster-users¶
Use this option to sync proxysql instances that manage multiple clusters.
This option does the following:
-
Syncs the currently configured Percona XtraDB Cluster user accounts with the ProxySQL database except for user accounts without a password and admin user accounts.
-
Keeps ProxySQL users that are not present in the Percona XtraDB Cluster
To sync a specific server combine this option with the --server option.
–-syncusers¶
This option does the following:
-
Syncs the currently configured Percona XtraDB Cluster user accounts with the ProxySQL database except for user accounts without a password and admin user accounts.
-
Deletes ProxySQL user accounts that are not also in Percona XtraDB Cluster from the ProxySQL database.
To sync a specific server combine this option with the --server option.
Review the user accounts in the ProxySQL database as root from ProxySQL database.
--syncusers example¶
proxysql admin#> SELECT DISTINCT username FROM mysql_users;
Expected output
+----------+
| username |
+----------+
| monitor |
+----------+
1 row in set (0.00 sec)
On a Percona XtraDB Cluster node, verify if a user account is already added.
mysql> SELECT user FROM mysql.user WHERE user LIKE 'test%';
Expected output
Empty set (0.00 sec)
On the node, add a new user.
mysql> CREATE USER 'test_user'@'localhost' IDENTIFIED WITH 'mysql_native_password' by 'passw0Rd';
Expected output
Query OK, 0 rows affected (0.04 sec)
Run percona-scheduler-admin
with the --syncusers
option:
$ percona-scheduler-admin --config-file=config.toml --syncusers
Expected output
Syncing user accounts from PXC(192.168.56.32:3306) to ProxySQL
Adding user to ProxySQL: test_user
Synced PXC users to the ProxySQL database!
Verify, in the ProxySQL database, that the user account has been added.
proxysql-admin> SELECT DISTINCT username FROM mysql_users;
Expected output
+-----------+
| username |
+-----------+
| monitor |
| test_user |
+-----------+
2 rows in set (0.00 sec)
--trace¶
This option enables shell-level tracing for this shell script.
–-update-cluster¶
This option checks the Percona XtraDB Cluster for new nodes. If nodes are found, they are added to ProxySQL. By default, offline nodes are not removed from the cluster.
Combining --remove-all-servers
with this option removes the server list
for the configuration before the update runs.
Combining --write-node
with this option, and if the node is in the
server list and is ONLINE
, then this command makes that node the
writer node by assigning a larger weight to that node. Only use this combination if the mode is singlewrite
.
$ percona-scheduler-admin --config-file=config.toml --write-node=127.0.0.1:4130 --update-cluster
Expected output
No new nodes detected.
Waiting for ProxySQL to process the new nodes...
Cluster node info
+---------------+-------+---------------+------+--------+---------+
| hostgroup | hg_id | hostname | port | status | weight |
+---------------+-------+---------------+------+--------+---------+
| writer | 100 | 192.168.56.34 | 3306 | ONLINE | 1000000 |
| reader | 101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.34 | 3306 | ONLINE | 1000000 |
| reader-config | 8101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
+---------------+-------+---------------+------+--------+---------+
Cluster membership updated in the ProxySQL database!
–-update-mysql-version¶
This option updates the mysql server version in the proxysql db based on the online writer node.
--update-mysql-version example¶
$ percona-scheduler-admin --config-file=config.toml --update-mysql-version
Expected output
ProxySQL MySQL version changed to 8.0.27
–-update-read-weight¶
Combining –update-cluster with this option assigns the specified read weight to a node.
The syntax for the arguments are:
--update-read-weight example¶
$ percona-scheduler-admin --config-file=config.toml --update-cluster --update-read-weight="<IP_ADDRESS:PORT>, <New Weight>"
The following table displays the pxc_scheduler_handler
default configuration for ProxySQL.
Cluster node info
+---------------+-------+---------------+------+--------+--------+
| hostgroup | hg_id | hostname | port | status | weight |
+---------------+-------+---------------+------+--------+--------+
| writer | 100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
+---------------+-------+---------------+------+--------+--------+
Cluster membership updated in the ProxySQL database!
The following command assigns the weight value of 1111
to the
192.168.56.32:3306
node in the reader and reader-config hostgroups.
$ percona-scheduler-admin --config-file=config.toml --update-cluster --update-read-weight="192.168.56.32:3306,1111"
Expected output
No new nodes detected.
Waiting for scheduler script to process the nodes...
Cluster node info
+---------------+-------+---------------+------+--------+--------+
| hostgroup | hg_id | hostname | port | status | weight |
+---------------+-------+---------------+------+--------+--------+
| writer | 100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.32 | 3306 | ONLINE | 1111 |
| writer-config | 8100 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.32 | 3306 | ONLINE | 1111 |
+---------------+-------+---------------+------+--------+--------+
Cluster membership updated in the ProxySQL database!
–-update-write-weight¶
Combining this option with --update-cluster assigns the specified write weight to a node.
The syntax for the arguments are:
--update-write-weight example¶
$ percona-scheduler-admin --config-file=config.toml --update-cluster --update-write-weight="<IP_ADDRESS:PORT>, <New Weight>"
The following table displays the pxc-scheduler-handler
default configuration for ProxySQL.
Cluster node info
+---------------+-------+---------------+------+--------+--------+
| hostgroup | hg_id | hostname | port | status | weight |
+---------------+-------+---------------+------+--------+--------+
| writer | 100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
+---------------+-------+---------------+------+--------+--------+
Cluster membership updated in the ProxySQL database!
The following command assigns the weight value of 1111
to the
192.168.56.33:3306
node in the writer and writer-config hostgroups.
$ percona-scheduler-admin --config-file=config.toml --update-cluster --update-write-weight="192.168.56.33:3306,1111"
Expected output
No new nodes detected.
Waiting for scheduler script to process the nodes...
Cluster node info
+---------------+-------+---------------+------+--------+--------+
| hostgroup | hg_id | hostname | port | status | weight |
+---------------+-------+---------------+------+--------+--------+
| writer | 100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader | 101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.34 | 3306 | ONLINE | 1000 |
| writer-config | 8100 | 192.168.56.33 | 3306 | ONLINE | 1111 |
| reader-config | 8101 | 192.168.56.32 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.33 | 3306 | ONLINE | 1000 |
| reader-config | 8101 | 192.168.56.34 | 3306 | ONLINE | 1000 |
+---------------+-------+---------------+------+--------+--------+
Cluster membership updated in the ProxySQL database!
--use-stdin-for-credentials¶
This option uses stdin
to send credentials to the MySQL client instead of process substitution. The default setting disables the option and uses process substitution.
--version¶
This option prints the version information.
–-write-node¶
This option chooses which Percona XtraDB Cluster node is the writer
node when the mode is singlewrite
. You can combine this option with –enable / -e and –update-cluster.
The option requires only a single IP address and port combination.
Assigning a node with --write-node gives the writer node a weight of 1000000. The default weight is 1000.
--write-node example¶
The argument syntax is a single IP address and port combination.
$ percona-scheduler-admin --config-file=config.toml --write-node=192.168.56.32:3306