Distributed Transaction database product version V2.5.2

Management explain the new features

Physical deployment environment

Context : How do I know built HotDB Server cluster operating environment eligibility, guidance on how to optimize the operation and maintenance personnel reasonably cluster operating environment to improve overall system performance? "Deployment environment health check" function provides reference for such issues
illustrate features :

Examination initiated task to be fulfilled clusters: cluster all component programs available SSH server configuration information, and the root is associated with the user, the normal operation of the cluster, the cluster configuration data at least a logical node and a library
Examination score out of 100, did not meet then deduct the points, but there is no penalty warning items specific recommendations or instructions, qualified item no special instructions and no points nor points
Environmental examination subdivision close to the 140 (the number of medical projects in different clusters are different modes), covering a cluster of hardware resources, operating systems, software deployment, software configuration, such as the five basic dimensions of functional verification
Support online viewing or export examination results
Physical examination results are divided into: non-compliance, warning items, qualified key, the default display items did not meet with the warning
Function:
Details page overview of examination results by the examination results panel with detailed physical examination of two parts
Whether there is a physical item or not reached quickly alert the dimension is determined by the dimension of each dot of red or orange in the upper right corner
After each dimension expands to determine whether there is a subdivision of the physical examination did not meet or alert the medical term basis under medical items tab label identifies the top right corner of the red or orange
Detailed examination results can be displayed for each subdivision of the physical examination: physical examination items, examination results, points, the exception object, tips, and other information criterion
Physical function deployment environment server version is not required, but lower than the physical version 2.5.2 is not supported "10 seconds Performance test", "standby configuration repository / MGR identity" two physical item detection
Cluster running information collection

Context :

HotDB Server cluster components are more operational mechanism is more complex. When abnormal problem or fault troubleshooting more difficult. The use of information-gathering tool for quick analysis of anomalies in the collection of the required log and configuration files cluster occurs when a user issues, thereby increasing the speed reducing troubleshooting time abnormal issues affecting the business.

Function Description:

Information collected is divided according to the scene to collect: different "health cluster", "performance test" in two  cluster model the collection list will be different
Part of the information gathering operations that require user authorization to open
Notes initiate collection task: to initiate the same time only one available SSH server configuration information (non-essential), compute nodes running normal (non-essential)
It collects information automatically downloaded to the local file package
Function:
Record page to view the task being initiated to gather information and historical task record
End of file collection stored in the default management platform for server / opt directory
History collection records provide file download function, but the corresponding files are deleted within the server can not be downloaded
Supports setting file window period, default 30 days, the program zero and trigger action to delete outdated files when you save the file deletion window period corresponding to the collection server when you delete the history every day

License Exception Monitoring remind
Function:
Support authorization expired license, permit detection of abnormalities, abnormal license information can not be updated three kinds of reminders
Remind mode: Web reminder (login or refreshed home when pop, top information bar compute nodes warning display), e-mail alert (you need to configure notification policies), compute node logs, compute node management side show @@ usbkey command
License information could not be updated to detect anomalies and will not affect the normal operation of computing nodes, compute nodes may be timely to restart service in off-peak business to solve the problem

- the configuration menu to optimize ease of use
Node Manager : the original data nodes, storage two secondary menu node comprising: a data management node, switching rules, from the main building, pause in the heartbeat, a storage node managing the storage node group management function blend six management functions for a node

node Manager:
adding data in the same node and the storage node " Add node "menu,
may be added to the data node by way of the storage node generated rapidly.
Storage nodes may be added separately.

Note:
① adding good data from the master node is automatically set up by default, automatically adapts high availability switching rules
② When the data from the node with a double master replication architecture, standby storage node may specify "the main storage node" field at the slave primary, standby or automatically match the storage node type is "master library" as the main
more configurable parameters ③ openable storage node set in the "operation" column [more] button

Table **: ** add bulk to a single table to add information to a unified "Add table information"
Here Insert Picture Description
NOTE:
Add single or batch number is determined by the input table name, batch add table is the table name requested by a newline, comma, English space be spaced in any manner. Fragmentation field interval with the colon.
Fragmentation rules :
new fragmentation rules can be automatically generated component sheet rule name, generation format: fragmentation rules ID + slice type name
MATCH fragmentation rules automatically increase setting mode, and the default blank string automatic sheet routing node
automatically or manually way to support a plurality of input values when the value of the specified range, comma or space interval
Here Insert Picture Description

Information backup failure logic optimization
Context:
direct construction of the table DDL (CREATE \ ALTER \ DROP \ TRUNCATE \ RENAME) to perform the operation table, might be relevant logic by automatically build tables grammar or logic library set the default data nodes to bypass management platform configuration repository affected, thus affecting the effectiveness of logical backup files, such as the right example.
Function Description:
hotdb_config new hotdb_ddl_log table, DDL for users to perform specific operations on compute nodes record:
specific DDL: Table bypass management platform built directly
modify, delete tables created by direct construction of the table
If it is found DDL statement to execute the backup file is automatically associated to the logical library library-related configuration logic influential set to fail
When starting the management platform, the management platform to run every hour, when performing a data recovery timing is determined whether three backup file needs to be set to invalid
Example:
Step 1, a default logic library fragments node
create table customer1 (id int (20 ) Not null AUTO_INCREMENT Primary Key, name VARCHAR (20 is));
INSERT INTO Customer1 values (nULL, 'A');
INSERT INTO Customer1 values (nULL, 'A');
step 2, a complete backup of
step 3, drop table if EXISTS Customer1;
Create Table Customer2 (ID int (20 is) Not null AUTO_INCREMENT Primary Key, name VARCHAR (20 is));
INSERT INTO Customer2 values (nULL, 'A');
INSERT INTO Customer2 values (nULL, 'A');
step 4, the first incremental backup
step 5, create table customer1 (id int (20) not null auto_increment primary key, name varchar (20));
step 6, a second incremental backup
step 7, drop table if exists customer1 ;
step 8, take a second to do incremental recovery
Will find customer1 does not appear by HotDB Server show tables, but will exist in the storage node. Because customer1 configuration has been removed in step 7;
From the master perfect build process
optimization build process:
new copy relationship is determined
to build a plurality of storage node instance is not target replication relationship, examples of the storage node is not set up with a plurality of storage node
copy relationship exists point
add duplication parameter detected
storage node instance structures correct parameters (corresponding to the binlog replication or replication parameters correctly opened GTID)
optimized master match detection data from
the detection data consistency table (except MySQL, INFORMATION_SCHEMA,
performance_schema, SYS library), if it is determined whether or not there is copy GTID off GTID point of
the new traffic detection
does not allow traffic on the storage node structures, or the failure to build
the new building is completed delay detection
automatically determines the data to build a complete master copy node relationship if there is a delay, if there is a reminder but does not affect the result of building

Note
standby state of the new "built in", "build failed" two states, set up during the leave or Log does not affect the background tasks to build the program. Failure to build support viewing record failure information.

- Additional features to optimize
the functional description:

Global optimization logic table data detection:
when the second detection task initiated when there is an overlap of the first table, if the second task is initiated by manually cancel cause failure or abnormal condition, did not affect the detection result of the first effectiveness (normally data repair)
increase the task is forcibly canceled entry
Standby consistency check:
manually cancel the execution of the tasks will keep a record display
Check the correctness of the routing data:
increasing the task "force cancel" operation, and displays the tasks manually cancel the record
Management platform for the functional specification page input box value
increased specification check box to enter the value of the input, increased checking on whether to allow the importation of Chinese, values, rules and other positive integers
**

- Server new features explained

- limit under XA mode for MySQL version
Context:
V5.7.17 following versions of MySQL support in XA matters is not stable enough, there are some official unrepaired defects. So when HotDB XA Transaction Server use MySQL version will be limited: MySQL version must be 5.7.17 and above.
Function:
In the open XA mode constraint behavior HotDB Server include:

If the start HotDB Server, detecting the presence of any MySQL version 5.7.17 below the storage node, HotDB Server fails to start;
2. If after starting HotDB Server, MySQL version is less than the storage node is added to 5.7.17, dynamic loading fails ;
If the pre-start HotDBServer, MySQL version 5.7.17 below the storage node can not connect, even after the re-start a successful connection, the storage node is still unavailable and dynamic loading will fail.

The above will output ERROR level log Tip: XA Rate this page Currently in the MODE, MySQL Version IS not
allowed to BE Lower Within last 5.7.17.
Time Zone Support Global
Context:
To ensure the accuracy of the data, address the time zone which led to inconsistent data inconsistencies Table insert timestamp data type, HotDB Server provides support for global time zones
Function:
When time_zone specific parameter values or the same and are all SYSTEM system_time_zone are all the same specific value, HotDBServer no special treatment;
Otherwise HotDB Server will be unified time_zone set to a fixed value: "+ 8:00" and will record a warning level log (The datasources' time_zones are not consistent), regardless of the value of the storage node time_zone set;
When backup and recovery, HotDB Server can guarantee recovery time before the backup is consistent with the type of data

DROP table data protection is not lost

Context:
Considering the DROP TABLE statement there is a higher risk in the actual production line environment or execution, HotDB Server supports table data protection DROP period of time is not lost.
Function Description:

Modify the parameters in server.xml turn this feature on, the default is not open.
dropTableRetentionTime = 0 (the default), represents not retained by DROP table, DROP TABLE i.e. drop table;
dropTableRetentionTime> 0, the unit is in hours, reservation is DROP table to the setting length, automatically delete the table the duration exceeds the set.

Function Description:
HotDB Server will retain the deleted data table, but it would not provide a direct restore operation command, when dropped table configuration is not made any changes, the user can restore the table on the storage node (note: this risk) : '
The first step, check the mapping between;
'
Table created by the configuration management platform that can RENAME directly into the original table; tables created by the automatic construction of the table function, the actual configuration has been deleted and can not be directly RENAME to the original table name.
After the reduction to automatically create a table that can be added to create a table on a management platform into the exemplar RENAME name.
After fragmentation rules table referenced by the deleted if they are modified, again reducing the table to be deleted, and the data will be consistent with the routing data before deletion, i.e., actual routing of data will not match the routing configuration changes.
If they are outside the presence of health or delete trigger on the table, deletes foreign key trigger retention and temporary tables. His son because of the foreign key table is deleted, the relationship between father and son after reduction is no longer the table.
o The third step is dynamically loaded into effect - Note: After you configure the database changes take effect dynamically loaded
Globally Unique Constraints
Context:
HotDB Server (V2.5.2 and below) can not guarantee that in some cases the column has a unique constraint on all nodes are unique, such as: The only constraint column is not a fragmented field. Now supports unique constraint globally unique in distributed.
Function Description:
GlobalUniqueConstraint parameter set to true (off by default), openable globally unique constraint
Note: this function enabled, SQL statements may INSERT, UPDATE, DELETE have a greater impact efficiency, and may result in increased operational delay SQL, and lock wait increase deadlock situation.
Another related parameters routeByRelativeCol (off by default), may be controlled: when the fragment does not contain the query statement SELECT field but contains only constraint field, query the secondary index to locate the fixed node, the SELECT query issued only under to the secondary index points to a node rather than a full table scan
For example: a conventional level slice table table01, fragmentation field id, fragmentation rules auto_mod, when executing the following query: SELECT * FROM table01 WHERE unique_col = 100; // unique_col this column is the only constraint only query issued to unique_col = data that a node 100, instead of all the data nodes
Mysqlbinlog supports syntax
Context :
HotDB Server supports incremental data synchronization syntax parsing mysqlbinlog to minimize downtime when migrating data to a single MySQL HotDB Server
Function Description :
The method using mysqlbinlog
first step, log in to the management port, perform dbremapping command to add database mapping relationship
dbremapping @@ add @ expected to be introduced into the database name: the logical name of the library
dbremapping @@ remove @ db01: db02 delete the mapping between
the second step, use mysqlbinlog binlog statement is executed in a selected SQL statement, using the following syntax and parameters required
mysqlbinlog --base64-output = decode-rows --skip-gtids --to-last-log --stop-never --database = database name --start-position = binlog starting position --stop-position = binlog terminating position binlog filename | mysql -u username -p password -h -P computing node server service port -c -A
e.g. desirable to 192.168 .200.77: 3306 in the physical library database db01 logicdb01 logic configured introduced on the management platform, the computing node belongs to the logical library 192.168.210.30.
Xianzhi 192.168.210.30 log in to the management port 3325, execute: dbremapping @@ add @ db01: logicdb01
192.168.200.77 to the directory then is performed as follows binlog command
mysqlbinlog --base64-output = decode-rows --skip-gtids --to-last-log --stop-never --database = db01 --start-position = 0 --stop-position = 190326 mysql- bin.000009 | mysql -uroot -proot -h192.168.210.30 -P3323 -c -A
HINT statement Special Instructions
The operational level is not recommended HINT direct operation of the storage node. The use of HINT, recommended the implementation of reconstruction after the completion of the new front-end connected
back-end operations are only allowed to bind the connection range after performing HINT operation, HotDB binds associated back-end logic library currently HINT query using the connection.
If connected to the tip when the reconstruction is performed subsequent HINT other operations involving data nodes outside binding, before binding the rear end of the connection will fail, the front end of the connection will be automatically reconstructed.
In addition HINT, the following statement will bind backend connector
set [session] foreign_key_checks = 0; // bind all the data nodes
START TRANSACTION / * 40100 WITH CONSISTENT SNAPSHOT * / // bindings associated logical bank currently in use! data node
set [session] UNIQUE_CHECKS = 0; // bind all the data nodes
Increasing distal warning / info and log information output
after executing the SQL statement HINT binds or other connection, the distal end will have a log of output warning information message info:
switching of logic library data associated with a different node, the front end output info message:
if when the back-end bindings connection is closed, the output of reminders and log information is being rebuilt at this time HotDB Server front-end connection:
Automatic adaptation startup processors, Executor
Context:
can set the following parameters for the current server configuration automatically adapts according to the rules when HotDB Server starts, in order to reduce the number of threads situation due to the problem of insufficient cause abnormal HotDB Server service:
16 4 4 Function :
Automatic adaptation rules:
Log 3325 port, performing show @@ threadpool; command values of the parameters
number following commands to display the current logical CPU:
CAT / proc / cpuinfo | grep "Processor" | -l WC
Note: maximum value of auto-adaptation
The introduction of NDB Cluster-SQL node Service
Context:
V2.5.2 version of the following HotDB Server to SQL statement SELECT query type of support is not complete, especially sub-types of SQL queries, so the introduction of NDB Cluster service. The services use the NDB Cluster SQL node characteristics compatible HotDB do not support SQL, used to complete relatively complex queries in a distributed computing environment.
After NDB SQL script installed successfully, to open the NDB service, the need to HotDB
ndbsqlmode modify the Server.xml Server configuration file is local.

Deployment instructions:
Deployment method is recommended to use the cluster management platform deployment capabilities, opens 'installation NDB SQL Services' in the 'cluster deployment'> 'parameters' in
After Ruoyi deploy compute nodes successful management platform, you can also append the script and then deploy the NDB SQL:
Log on to execute a script on each compute node server. Note: Each host server compute nodes
are required to deploy the NDB SQL.
Enter key to deploying a default installation directory: cd / usr / local / hotdb / Install_Package, execute: sh hotdbinstall_beta.sh --install-ndbsql = yes --ntpdate-server-host = 192.168.210.221
Note: Recommended NDB SQL-time follower HotDB Server deployment is successful, if the deployment of additional follow-up, the need to pay strict attention to -install-ntpd, -ntpdate-server- ip, -ntpdate-server-host mass participation problem that requires synchronization configuration consistent with the current time

function:
NDB SQL node does not need a separate startup or shutdown, start HotDB Server will be started automatically
NDB SQL node, closed versa.
When NDB the SQL mode is enabled, provided the use NDB logic is to meet the following conditions:
① the case of a globally unique not open, must have a primary key or a unique key, primary and unique keys must be incremented for the fragment field or;
② globally unique open under the circumstances, you must have a primary key or a unique key;
③ primary key the two pieces of the unique key must be a single field numeric type;
④ performed HotDB original unsupported query
such as: SELECT ... FROM ... / LEFT JOIN / RIGHT JOIN ... ON ... WHERE condition may be used behind the equal sign (<=> or <>), or take the remainder quotient (only support column% div constant or constant) and the like;
the SELECT ... ... the FROM / the JOIN the LEFT / RIGHT behind the JOIN ... ON condition can with iN / iS NOT NULL / iS NULL / BETWEEN ... aND / lIKE, inequality sign (<=> or <>), XOR the like;
three tables above and other multi-table queries;
NDB restrictions:
table structure if the non-geometry type of space type field, json type of field, the original does not support SQL queries still can not support;
the WHERE conditions of the original does not support SQL queries in the field type supports only: All plastic, decimal, char, enum, varchar type.
- Special Instructions
a key deployment package update instructions
The default installation of MySQL 5.7 version 5.7.23 and replaced with MySQL 5.7.25
reason: MySQL 5.7.23 version has now found a serious bug, see https://bugs.mysql.com/bug.php?id=92108