Set up Replication on HA Pair (4.7 and Below)

Customer Managed Applies to customer-managed instances of Alation


  • The same Alation version must be installed on both servers.

  • An Alation Data disk and Backup disk must be configured on the Secondary.

  • Verify that the Alation Data and Backup disk are configured by running one of the following commands, depending on Alation version.

    • For versions 3.12 and below:

      tail -n +1 /usr/local/alation/.disk*

    /usr/local/alation/.disk1_cache path is /data

    /usr/local/alation/.disk2_cache path is /backup

    • For versions 4.0 and above:

      tail -n +1 /opt/alation/alation/.disk*

    /usr/local/alation/.disk1_cache path is /data

    /usr/local/alation/.disk2_cache path is /backup

    If .disk1_cache or .disk2_cache have invalid paths, run the following command on the Secondary:

    sudo/etc/init.d/alation init DATA_DISK_PATH BACKUP_DISK_PATH
  • Alation Services need to be started up on both servers. If Alation is not started on both servers run the following command:

    sudo/etc/init.d/alation start
  • Alation Must be Configured as standalone mode on both servers

    sudo /etc/init.d/alation shell
    alation_conf alation.cluster.replication_mode
    alation.cluster.replication_mode = standalone
  • If the server is not in standalone mode, run the following command on both servers:

    sudo /etc/init.d/alation shell
    alation_action cluster_enter_standalone_mode

The Primary and Secondary must be able to communicate to each other over the following ports:

Port Number

Port Function


Postgres Replication


MongoDB Replication




SSH, used for signaling & file transfers between servers

Replication Setup Procedures

Step 1:  Collect System Details

To get the ip address with CIDR try running ip addr from the host.

If Multiple NICS are detected select the one used to access the server:

(env) PROD [alation @ip-10-11-2-15 /]$ ip addr
lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc no queue state UNKNOWN
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet scope host lo
    inet6 ::1/128 scope host
      valid_lft forever preferred_lft forever

eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9001 qdisc pfifo_fast state UP qlen 1000
    link/ether 12:e5:2b:d0:eb:8b brd ff:ff:ff:ff:ff:ff
    inet brd scope global eth0
    inet6 fe80::10e5:2bff:fed0:eb8b/64 scope link
      valid_lft forever preferred_lft forever

To get PostgreSQL disk usage, try running the following Alation shell command:

(env) PROD {alation@ip-10-11-2-15 /j$ sudo du -hs /data1/pgsql/9.3/data/ 12G /data1/pgsql/9.3/data/

Use the following chart to fill in the required information.


Address with CIDR

SSH Username, default: alationadmin





Step 2:  Key Exchange

To get HA you need passwordless SSH between the Primary ↔ Secondary. Alation ships a python script to help set this up. The script should be run on the host outside the chroot.

The script creates only an ssh key pair and place it into authorized_keys for the Alation admin user. If you want to use another user, which should not occur often, create the authorized_keys manually for that user.

  1. On the Primary, use the Secondary IP as the command line argument.

For versions 3.12 and below:

sudo python /usr/local/alation/opt/alation/ops/actions/unjailed_key_exchange

For versions 4.0 and above:

sudo python /opt/alation/alation/opt/alation/ops/actions/unjailed_key_exchange
  1. On the Secondary, use the Primary IP as the command line argument.

For versions 3.12 and below:

sudo python /usr/local/alation/opt/alation/ops/actions/unjailed_key_exchange

For versions 4.0 and above:

sudo python /opt/alation/alation/opt/alation/ops/actions/unjailed_key_exchange

The script creates replication_rsa key files on the Secondary and Primary. You are asked to copy over the public key from Secondary to Primary and Primary to Secondary.

Step 3:  Verify SSH

  1. From the chroot shell on the Primary, SSH to the Secondary. After you have logged in, it is a good idea to touch .hushlogin to prevent any MOTD from causing errors later.

  2. From the Primary:

ssh -i/home/alation/.ssh/replication_rsa -l alationadmin touch .hushlogin
  1. From the Secondary:

ssh -i/home/alation/.ssh/replication_rsa -l alationadmin touch .hushlogin

Step 4:  Change Host from Standalone to Primary Mode

  1. The alation_action command must be executed inside the Alation shell. Alation shell is entered by running:

sudo/etc/init.d/alation shell
  1. From the Primary, run:

alation_action cluster_enter_master_mode

Step 5:  Replicate to Secondary

  1. The alation_conf command must be executed inside the Alation shell. Alation shell is entered by running:

sudo/etc/init.d/alation shell
  1. From the Primary, add the Secondary IP CIDR to alation_conf:

alation_conf alation.cluster.hosts -s['']
  1. Decide if we are going to attempt PostgreSQL replication in-line. General guidance:

  • Do not try to replicate PostgreSQL in-line for installs > 100GB PostgreSQL data files

  • If downtime needs to be minimized, do not replicate PostgreSQL in-line.

  • If new install with minor data, replicating in-line will be OK.

If you choose not to perform replication in-line, continue to Step 5a.

  1. From the Primary, add the second host as Secondary:

alation_action cluster_add_slaves
  1. If you choose not to perform replication in-line, go to step 5b.

Step 5a: Disable PostgreSQL In-line Replication

  1. From the Secondary, set alation_conf item to not replicate postgres in-line:

alation_conf alation.cluster.pg_repl_in_line -s False
  1. Continue to Step 6.

Step 5b:  Replicate PostgreSQL

From the Secondary, replicate postgres using the alation_action command:

alation_action cluster_replicate_postgres

Step 6:  Verification

  1. Verify PostgreSQL Replication status.

    The following query shows the Secondary instances that are attached to the Primary and any replication delay, in bytes:

    psql rosemeta
    SELECT client_hostname, client_addr,
    pg_stat_replication.replay_location) AS byte_lag FROM pg_stat_replication;
  2. Verify Mongo Lag.

    The following query displays the number of seconds the Secondary lags behind the Primary. It is expected for this verification to return a large number of hours when replication is the first setup. It eventually works through the backlog and catches up.