Configuring Client Authentication
Configuring Client Authentication
Describes the available methods for authenticating Greenplum Database clients.
Allowing Connections to Greenplum Database
Client access and authentication is controlled by a configuration file named pg_hba.conf (the standard PostgreSQL host-based authentication file). For detailed information about this file, see The pg_hba.conf File in the PostgreSQL documentation.
In Greenplum Database, the pg_hba.conf file of the master instance controls client access and authentication to your Greenplum system. The segments also have pg_hba.conf files, but these are already correctly configured to only allow client connections from the master host. The segments never accept outside client connections, so there is no need to alter the pg_hba.conf file on segments.
host database role address authentication-method
local database role authentication-method
- Matches connection attempts using UNIX-domain sockets. Without a record of this type, UNIX-domain socket connections are disallowed.
- Matches connection attempts made using TCP/IP. Remote TCP/IP connections will not be possible unless the server is started with an appropriate value for the listen_addresses server configuration parameter.
- Matches connection attempts made using TCP/IP, but only when the connection is made with SSL encryption. SSL must be enabled at server start time by setting the ssl configuration parameter. Requires SSL authentication be configured in postgresql.conf. See Configuring postgresql.conf for SSL Authentication.
- Matches connection attempts made over TCP/IP that do not use SSL. Requires SSL authentication be configured in postgresql.conf. See Configuring postgresql.conf for SSL Authentication.
- Specifies which database names this record matches. The value all specifies that it matches all databases. Multiple database names can be supplied by separating them with commas. A separate file containing database names can be specified by preceding the file name with @.
- Specifies which database role names this record matches. The value all specifies that it matches all roles. If the specified role is a group and you want all members of that group to be included, precede the role name with a +. Multiple role names can be supplied by separating them with commas. A separate file containing role names can be specified by preceding the file name with a @.
- Specifies the client machine addresses that this record matches. This field can contain an IP address, an IP address range, or a host name.
- An IP address range is specified using standard numeric notation for the range's starting address, then a slash (/) and a CIDR mask length. The mask length indicates the number of high-order bits of the client IP address that must match. Bits to the right of this should be zero in the given IP address. There must not be any white space between the IP address, the /, and the CIDR mask length.
- Typical examples of an IPv4 address range specified this way are 172.20.143.89/32 for a single host, or 172.20.143.0/24 for a small network, or 10.6.0.0/16 for a larger one. An IPv6 address range might look like ::1/128 for a single host (in this case the IPv6 loopback address) or fe80::7a31:c1ff:0000:0000/96 for a small network. 0.0.0.0/0 represents all IPv4 addresses, and ::0/0 represents all IPv6 addresses. To specify a single host, use a mask length of 32 for IPv4 or 128 for IPv6. In a network address, do not omit trailing zeroes.
- An entry given in IPv4 format will match only IPv4 connections, and an entry given
in IPv6 format will match only IPv6 connections, even if the represented address is in
the IPv4-in-IPv6 range.
Note: Entries in IPv6 format will be rejected if the host system C library does not have support for IPv6 addresses.
- If a host name is specified (an address that is not an IP address or IP range is treated as a host name), that name is compared with the result of a reverse name resolution of the client IP address (for example, reverse DNS lookup, if DNS is used). Host name comparisons are case insensitive. If there is a match, then a forward name resolution (for example, forward DNS lookup) is performed on the host name to check whether any of the addresses it resolves to are equal to the client IP address. If both directions match, then the entry is considered to match.
- Some host name databases allow associating an IP address with multiple host names, but the operating system only returns one host name when asked to resolve an IP address. The host name that is used in pg_hba.conf must be the one that the address-to-name resolution of the client IP address returns, otherwise the line will not be considered a match.
- When host names are specified in pg_hba.conf, you should ensure that name resolution is reasonably fast. It can be of advantage to set up a local name resolution cache such as nscd. Also, you can enable the server configuration parameter log_hostname to see the client host name instead of the IP address in the log.
- These fields can be used as an alternative to the CIDR address notation. Instead of specifying the mask length, the actual mask is specified in a separate column. For example, 255.0.0.0 represents an IPv4 CIDR mask length of 8, and 255.255.255.255 represents a CIDR mask length of 32.
- Specifies the authentication method to use when connecting. See Authentication Methods for options.
Editing the pg_hba.conf File
Initially, the pg_hba.conf file is set up with generous permissions for the gpadmin user and no database access for other Greenplum Database roles. You will need to edit the pg_hba.conf file to enable users' access to databases and to secure the gpadmin user. Consider removing entries that have trust authentication, since they allow anyone with access to the server to connect with any role they choose. For local (UNIX socket) connections, use ident authentication, which requires the operating system user to match the role specified. For local and remote TCP connections, ident authentication requires the client's host to run an indent service. You could install an ident service on the master host and then use ident authentication for local TCP connections, for example 127.0.0.1/28. Using ident authentication for remote TCP connections is less secure because it requires you to trust the integrity of the ident service on the client's host.
This example shows how to edit the pg_hba.conf file of the master to allow remote client access to all databases from all roles using encrypted password authentication.
- Open the file $MASTER_DATA_DIRECTORY/pg_hba.conf in a text editor.
- Add a line to the file for each type of connection you want to allow. Records are read
sequentially, so the order of the records is significant. Typically, earlier records
will have tight connection match parameters and weaker authentication methods, while
later records will have looser match parameters and stronger authentication methods. For
# allow the gpadmin user local access to all databases # using ident authentication local all gpadmin ident sameuser host all gpadmin 127.0.0.1/32 ident host all gpadmin ::1/128 ident # allow the 'dba' role access to any database from any # host with IP address 192.168.x.x and use md5 encrypted # passwords to authenticate the user # Note that to use SHA-256 encryption, replace md5 with # password in the line below host all dba 192.168.0.0/32 md5
- Basic Authentication
- Kerberos Authentication
- LDAP Authentication
- SSL Client Authentication
- PAM Based Authentication
- Radius Authentication
- Password or MD5
- Requires clients to provide a password, one of either:
- Md5 – password transmitted as an MD5 hash.
- Password – A password transmitted in clear text. Always use SSL connections to prevent password sniffing during transit. This is configurable, see "Encrypting Passwords" in the Greenplum Database Administrator Guide for more information.
- Reject the connections with the matching parameters. You should typically use this to restrict access from specific hosts or insecure connections.
- Authenticates based on the client's operating system user name. This is secure for local socket connections. Using ident for TCP connections from remote hosts requires that the client's host is running an ident service. The ident authentication method should only be used with remote hosts on a trusted, closed network.
hostnossl all all 0.0.0.0 reject hostssl all testuser 0.0.0.0/0 md5 local all gpuser ident
You can authenticate against a Kerberos server (RFC 2743, 1964).
- Map system and database users.
- Option to specify realm name included in the system-user name in the ident map file.
- Specify the realm name for matching the principals.
- The hostname of the service principal.
host all all 0.0.0.0/0 krb5 hostssl all all 0.0.0.0/0 krb5 map=krbmap
- krb_server_key file
- Sets the location of the Kerberos server key file.
- krb_srvname string
- Kerberos service name.
- krb_caseins_users boolean
- Case-sensitivity. The default is off.
- The Kerberos service name to use for authentication.
You can authenticate against an LDAP directory.
- LDAPS and LDAP over TLS options encrypt the connection to the LDAP server.
- The connection from the client to the server is not encrypted unless SSL is enabled. Configure client connections to use SSL to encrypt connections from the client.
- To configure or customize LDAP settings, set the LDAPCONF environment variable with the path to the ldap.conf file and add this to the greenplum_path.sh script.
- Set up the LDAP server with the database users/roles to be authenticated via LDAP.
- On the database:
- Verify that the database users to be authenticated via LDAP exist on the database. LDAP is only used for verifying username/password pairs, so the roles should exist in the database.
- Update the pg_hba.conf file in the $MASTER_DATA_DIRECTORY to use LDAP as the authentication method for the respective users. Note that the first entry to match the user/role in the pg_hba.conf file will be used as the authentication mechanism, so the position of the entry in the file is important.
- Reload the server for the pg_hba.conf configuration settings to take effect (gpstop -u).
- Names or IP addresses of LDAP servers to connect to. Multiple servers may be specified, separated by spaces.
- String to prepend to the user name when forming the DN to bind as, when doing simple bind authentication.
- String to append to the user name when forming the DN to bind as, when doing simple bind authentication.
- Port number on LDAP server to connect to. If no port is specified, the LDAP library's default port setting will be used.
- Set to 1 to make the connection between PostgreSQL and the LDAP server use TLS encryption. Note that this only encrypts the traffic to the LDAP server — the connection to the client will still be unencrypted unless SSL is used.
- Root DN to begin the search for the user in, when doing search+bind authentication.
- DN of user to bind to the directory with to perform the search when doing search+bind authentication.
- Password for user to bind to the directory with to perform the search when doing search+bind authentication.
- Attribute to match against the user name in the search when doing search+bind authentication.
ldapserver=ldap.greenplum.com prefix="cn=" suffix=", dc=greenplum, dc=com"
host all testuser 0.0.0.0/0 ldap ldap ldapserver=ldapserver.greenplum.com ldapport=389 ldapprefix="cn=" ldapsuffix=",ou=people,dc=greenplum,dc=com" hostssl all ldaprole 0.0.0.0/0 ldap ldapserver=ldapserver.greenplum.com ldaptls=1 ldapprefix="cn=" ldapsuffix=",ou=people,dc=greenplum,dc=com"
SSL Client Authentication
SSL authentication compares the Common Name (cn) attribute of an SSL certificate provided by the connecting client during the SSL handshake to the requested database user name. The database user should exist in the database. A map file can be used for mapping between system and database user names.
SSL Authentication Parameters
- CertAuthentication options:
Following are sample pg_hba.conf entries for SSL client authentication:
- Connection type must be hostssl.
- This is specified in the pg_ident.conf, or in the file specified in the ident_file server setting.
Hostssl testdb certuser 192.168.0.0/16 cert Hostssl testdb all 192.168.0.0/16 cert map=gpuser
Greenplum Database reads the OpenSSL configuration file specified in $GP_HOME/etc/openssl.cnf by default. You can make changes to the default configuration for OpenSSL by modifying or updating this file and restarting the server.
Creating a Self-Signed Certificate
A self-signed certificate can be used for testing, but a certificate signed by a certificate authority (CA) (either one of the global CAs or a local one) should be used in production so that clients can verify the server's identity. If all the clients are local to the organization, using a local CA is recommended.
- Enter the following openssl
openssl req -new -text -out server.req
- Enter the requested information at the prompts.
Make sure you enter the local host name for the Common Name. The challenge password can be left blank.
- The program generates a key that is passphrase-protected; it does not accept a
passphrase that is less than four characters long. To remove the passphrase (and you
must if you want automatic start-up of the server), run the following
openssl rsa -in privkey.pem -out server.key rm privkey.pem
- Enter the old passphrase to unlock the existing key. Then run the following
openssl req -x509 -in server.req -text -key server.key -out server.crt
This turns the certificate into a self-signed certificate and copies the key and certificate to where the server will look for them.
- Finally, run the following
chmod og-rwx server.key
For more details on how to create your server private key and certificate, refer to the OpenSSL documentation.
Configuring postgresql.conf for SSL Authentication
- ssl boolean. Enables SSL connections.
- ssl_renegotiation_limit integer. Specifies the data limit before key renegotiation.
- ssl_ciphers string. Lists SSL ciphers that are allowed.
- server.crt. Server certificate.
- server.key. Server private key.
- root.crt. Trusted certificate authorities.
- root.crl. Certificates revoked by certificate authorities.
Configuring the SSL Client Connection
- Only use SSL connection. If a root CA file is present, verify the certificate in the same way as if verify-ca was specified.
- Only use an SSL connection. Verify that the server certificate is issued by a trusted CA.
- Only use an SSL connection. Verify that the server certificate is issued by a trusted CA and that the server host name matches that in the certificate.
- The file name of the client SSL certificate. The default is ~/.postgresql/postgresql.crt.
- The secret key used for the client certificate. The default is ~/.postgresql/postgresql.key.
- The name of a file containing SSL Certificate Authority certificate(s). The default is ~/.postgresql/root.crt.
- The name of the SSL certificate revocation list. The default is ~/.postgresql/root.crl.
- sslmode – PGSSLMODE
- sslkey – PGSSLKEY
- sslrootcert – PGSSLROOTCERT
- sslcert – PGSSLCERT
- sslcrl – PGSSLCRL
PAM Based Authentication
"PAM" (Pluggable Authentication Modules) is used to validate username/password pairs, similar to basic authentication. PAM authentication only works if the users already exist in the database.
- The default PAM service is postgresql. Note that if PAM is set up to read /etc/shadow, authentication will fail because the PostgreSQL server is started by a non-root user.
local all gpuser am pamservice=postgresql
RADIUS (Remote Authentication Dial In User Service) authentication works by sending an Access Request message of type 'Authenticate Only' to a configured RADIUS server. It includes parameters for user name, password (encrypted), and the Network Access Server (NAS) Identifier. The request is encrypted using the shared secret specified in the radiussecret option. The RADIUS server responds with either Access Accept or Access Reject.
RADIUS authentication only works if the users already exist in the database.
The RADIUS encryption vector requires SSL to be enabled in order to be cryptographically strong.
RADIUS Authentication Options
- The name of the RADIUS server.
- The RADIUS shared secret.
- The port to connect to on the RADIUS server.
- NAS identifier in RADIUS requests.
hostssl all all 0.0.0.0/0 radius radiusserver=servername radiussecret=sharedsecret
Limiting Concurrent Connections
To limit the number of active concurrent sessions to your Greenplum Database system, you can configure the max_connections server configuration parameter. This is a local parameter, meaning that you must set it in the postgresql.conf file of the master, the standby master, and each segment instance (primary and mirror). The value of max_connections on segments must be 5-10 times the value on the master.
When you set max_connections, you must also set the dependent parameter max_prepared_transactions. This value must be at least as large as the value of max_connections on the master, and segment instances should be set to the same value as the master.
- Stop your Greenplum Database system:
- On the master host, edit $MASTER_DATA_DIRECTORY/postgresql.conf and
change the following two parameters:
- max_connections – the number of active user sessions you want to allow plus the number of superuser_reserved_connections.
- max_prepared_transactions – must be greater than or equal to max_connections.
- On each segment instance, edit SEGMENT_DATA_DIRECTORY/postgresql.conf
and change the following two parameters:
- max_connections – must be 5-10 times the value on the master.
- max_prepared_transactions – must be equal to the value on the master.
- Restart your Greenplum Database system:
Encrypting Client/Server Connections
Greenplum Database has native support for SSL connections between the client and the master server. SSL connections prevent third parties from snooping on the packets, and also prevent man-in-the-middle attacks. SSL should be used whenever the client connection goes through an insecure link, and must be used whenever client certificate authentication is used.
To enable SSL requires that OpenSSL be installed on both the client and the master server systems. Greenplum can be started with SSL enabled by setting the server configuration parameter ssl=on in the master postgresql.conf. When starting in SSL mode, the server will look for the files server.key (server private key) and server.crt (server certificate) in the master data directory. These files must be set up correctly before an SSL-enabled Greenplum system can start.
A self-signed certificate can be used for testing, but a certificate signed by a certificate authority (CA) should be used in production, so the client can verify the identity of the server. Either a global or local CA can be used. If all the clients are local to the organization, a local CA is recommended. See Creating a Self-Signed Certificate for steps to create a self-signed certificate.