gs_initdb

Overview

Background

When a database is initialized using gs_initdb, a database directory is created, a system catalog is generated, and a default database and a template database are created.

System catalog

A large number of system catalogs and views are generated when a database is initialized, many of which allow access from any database user.

NOTE: Permissions on pg_user_status and pg_auth_history system catalogs are granted only to initialized users and sysadmin users.

Generated database

template1: a template database. When you create a database, all content in the template1 database is copied. The settings of template1 are determined by the parameters of gs_initdb.
template0: Is the initial backup database provided by MogDB. template0 can be used to generate an empty database if necessary.
postgres: a default database provided for users, tools, and third-party applications.

Usage Guide

Background

During installation, you are advised to use the -D parameter to call gs_initdb to initialize a database. If a database needs to be initialized to rectify a fault, run gs_initdb.

Although gs_initdb attempts to create the corresponding data directory, it may not have the permission to do so, because in most cases, the parent directory is owned by user root. To create a data directory, create an empty data directory as user root first and deliver the ownership of this directory to the database user using chown.
gs_initdb is used to set the template1 database and the setting becomes the default setting of other databases.
gs_initdb initializes the default locale and character set encoding of a database. The character set encoding, character encoding sequence (LC_COLLATE), and character set class (LC_CTYPE, such as uppercase letters, lowercase letters, and digits) can be separately set for databases when you create them.

syntax

gs_initdb [OPTION]... [DATADIR]

Procedure

Log in as the OS user omm to the primary database node.
Plan the database directory.

a. Switch to user root.
```
 su - root
```
b. Enter the username and password as prompted.
Go to the /opt/gaussdb directory and create the data1 directory.
```
cd /opt/gaussdb
mkdir data1
```
a. Allocate the ownership of the data1 directory to the database user omm. dbgrp is the user group that the omm belongs to.
```
chown omm:dbgrp data1
```
b. Exit user root.
```
exit
```

Run the gs_initdb command to initialize the database.

gs_initdb -D /opt/gaussdb/data1 -w "Enmo@123" --nodename='data1'

Command Reference

For details about common parameters and uncommon parameters supported by gs_initdb, see Table 1 and Table 2.

Table 1 Common parameters

Parameter	Description	Value Range
-A, --auth=METHOD	Specifies the authentication method for local users used in pg_hba.conf (host and local rows). Do not use trust, which is the default value, unless you trust all local users on your system. NOTICE: If the value is md5, manually change the value of password_encryption_type in the postgresql.conf.sample file to 0 and comment out the parameter for the change to take effect. gs_initdb must be used in conjunction with -W.	Values of METHOD: - trust - reject - md5 (an insecure algorithm, used for compatibility with earlier versions) - sha256 - sm3 Default value: trust
--auth-host=METHOD	Specifies the authentication method for local users over TCP/IP used in pg_hba.conf (host rows). The value specified in the -A parameter is overwritten when this parameter is specified.	Values of METHOD: - trust - reject - md5 (an insecure algorithm, used for compatibility with earlier versions) - sha256 - sm3 Default value: trust
--auth-local=METHOD	Specifies the authentication method for local users using Unix-domain socket connections used in pg_hba.conf (local rows). The value specified in the -A parameter is overwritten when this parameter is specified.	Values of METHOD: - trust - reject - md5 (an insecure algorithm, used for compatibility with earlier versions) - sha256 - sm3 Default value: trust
-c, --enable-dcf	Sets the node to be installed to the DCF mode.	-
[-D, --pgdata=]DATADIR	Specifies the data directory.	Set DATADIR as required. The value cannot contain the following characters: \|, ;, &, $, <, >, `, \, !
--nodename=NODENAME	Specifies the name of an initialized node.	The naming rules for nodes are as follows: - The node name can only be lowercase letters (a-z), underscores (_), number signs (#), digits (0-9), and dollar signs ($). - The node name must start with a lowercase letter (a-z), or underscore (_). - The nodename cannot be an empty string, and cannot exceed 64 characters.
-E, --encoding=ENCODING	Specifies the encoding format for a new database.	If this parameter is specified, the --locale option must be added to specify the locale that supports the encoding format. If the --locale option is not added, the system default locale is used. If the encoding format in the system default locale is inconsistent with the encoding format specified in this parameter, the database initialization failed. If this parameter is not specified, the encoding format of the system default locale is used. The system default locale and encoding format can be checked using the locale command as follows: omm@linux:~> locale\|grep LC_CTYPE LC_CTYPE="en_US.UTF-8" UTF-8 indicates the encoding format of the system default locale. The value cannot contain the following characters: \|, ;, &, $, <, >, `, \, !
--locale=LOCALE	Specifies the default locale for the new database. You can run the locale -a command to view available locales, for example, `zh_CN.gbk`. If you do not want to specify a locale, set the parameter to C. NOTICE: If the encoding format of the database is set, the encoding format of the user specified area must be consistent with the encoding format set by the user. Otherwise, database initialization fails.	For example, to set the database encoding format to GBK, perform the following steps: 1. Run the locale -a \|grep gbk command to check the locale that supports GBK encoding: omm@linux:~> locale -a\|grep gbk zh_CN.gbk zh_SG.gbk 2. Add the `--locale=zh_CN.gbk` option when the database is initialized.
--dbcompatibility=DBCOMPATIBILITY	Specifies the type of the compatible database.	Value range: A, B, C, and PG, indicating O, MY, TD, and POSTGRES databases, respectively.
--lc-collate=LOCALE --lc-ctype=LOCALE --lc-messages=LOCALE --lc-monetary=LOCALE --lc-numeric=LOCALE --lc-time=LOCALE	Sets the specified locale for a new database.	The parameter values must be supported by the OS. The value cannot contain the following characters: \|, ;, &, $, <, >, `, \, ! NOTE: If the --lc-collate parameter is not specified when the database is installed, the default value of --lc-collate is C.
--no-locale	Equivalent to --locale=C.	-
--pwfile=FILE	Reads the password for the database administrator from FILE during the running of gs_initdb. The first row of the file is taken as the password.	The format of FILE can be "a relative path+file" or "an absolute path+file". The relative path is relative to the current path. The value cannot contain the following characters: \|, ;, &, $, <, >, `, \, !
-T, --text-search-config=CFG	Specifies the default text search mode. The value of this parameter is not verified. After the configuration is successful, a log is recorded to notify you of the value of this parameter.	Values of text-search-config: - english: full-document search - simple: common text search Default value: simple
-U, --username=NAME	Selects the username of the database administrator.	Value range: normal database users. The value cannot contain the following characters: \|, ;, &, $, <, >, `, \, ! Default value: OS user who runs gs_initdb
-W, --pwprompt	Prompts users to enter the password for the database administrator during the running of gs_initdb.	-
-w, --pwpasswd=PASSWD	Specifies the password of the database administrator by running commands during the running of gs_initdb instead of interactive input.	The password must meet the following complexity requirements: - Contain at least eight characters. - Cannot be the same as the username, the current password (ALTER), or the current password in an inverted sequence. - Contain at least three of the following: uppercase characters, lowercase characters, digits, and special characters (limited to ~!@#$%^&*()-_=+\\|[{}];:,<.>/?). Note: The "$" symbol is a special symbol for referencing variables in the shell, so you need to add "\" to escape it, or use single quotes to wrap the password when executing in the terminal.
-C, --enpwdfiledir=DIR	Specifies the directory where the password encrypted is located, using the AES128 encryption algorithm during the running of gs_initdb. gs_initdb decrypts the password file in this directory and performs the password complexity check on the decrypted password. The password is used as the user's password if the check passes. NOTE: - The decrypted password file must use gs_guc to generate `gs_guc encrypt -K Enmo@123 -D Dir`. - If multiple -w and -C parameters are specified, gs_initdb regards the last -w or -C parameter as the user's requirement (entering a plain-text password or a password encrypted using AES128).	The value cannot contain the following characters: \|, ;, &, $, <, >, `, \, !
-X, --xlogdir=XLOGDIR	Specifies the directory where the transaction logs are stored. The directory must be a directory where an MogDB user has the read and write permissions.	The value must be an absolute path. The value cannot contain the following characters: \|, ;, &, $, <, >, `, \, !
-S, --security	Initializes a database in a secure mode.	After the database is initialized using -S, the created database user permissions are restricted, and the public schema permission cannot be used any more by default.

Table 2 Uncommon parameters

Parameter	Description	Value Range
-d, --debug	Prints debugging output from the bootstrap backend. The bootstrap backend is used by gs_initdb to create system catalogs.	-
-L DIRECTORY	Specifies the input file path where the database is initialized using gs_initdb. This is unnecessary. You will be told if you need to specify their location explicitly. This parameter is used to create a database with specified configuration information. You are advised to copy all subdirectories and files related to startup in the share/postgresql directory to avoid impact of other factors.	Path of input files required by the initialized database.
-n, --noclean	If this parameter is not specified and gs_initdb determines that an error prevents it from creating a database, it removes any files it might have created before discovering that it cannot finish the job. This option inhibits tidying-up and is useful for debugging.	-
-s, --show	Displays internal settings.	-
-V, --version	Prints the gs_initdb version and exits.	-
-?, --help	Displays help about gs_initdb command line parameters and exits.	-

Issue