Security & identity

ManageEngine PAM360 High Availability: Architecture, Setup Steps and Port Requirements

Choose the right PAM360 high availability model, follow the documented PostgreSQL and SQL Server setup steps, open the correct ports and confirm in the console that HA is alive.

By EpiFive Editorial TeamReviewed/updated 13 min readHow we check our guides
On this page

PAM360 high availability means running more than one PAM360 server so that privileged passwords stay reachable when the primary server or its database fails. ManageEngine documents several models: a primary and secondary pair on a shared database, an added read-only server, and a Microsoft SQL Server secondary model. The steps below follow ManageEngine's help pages as of 2026-11-20, and they start with choosing the model, because the build number and the database decide which guide applies.

Who this is for and what you will have at the end

This guide is for administrators who run, or are about to deploy, ManageEngine PAM360 and need continuity if a server goes down. At the end you will know which HA model fits your database, which ports to open, the order of the setup steps for the PostgreSQL and SQL Server paths, and where the console confirms that HA is running.

It is a planning and orientation guide that follows the vendor documentation. It does not replace the step pages for your build, and it deliberately skips values the pages do not state, such as licensing terms.

The PAM360 HA and DR models

ManageEngine's "High Availability and Disaster Recovery Models in PAM360" page groups the options like this:

ModelDatabase in the guide titlesWhat it does
Failover ServiceMicrosoft SQL Server clustersMoves operations to a standby server when the primary fails, without manual steps
Read-Only ServerPostgreSQLRead-only servers stay in sync with the primary and can take over if it fails
Application ScalingPostgreSQL and Microsoft SQL ServerSpreads load across clusters to improve performance and availability
Secondary Server ArchitecturePostgreSQL and Microsoft SQL ServerSecondary servers replicate the primary's state to serve as standbys

The same page says backup and restore (disaster recovery) works for both databases: scheduled backups, secure backup storage and restoration after data loss. Treat backups as separate from HA. A replicated database copies a bad change as faithfully as a good one, so a scheduled backup remains your way back.

The page also notes that read-only servers can be configured in several locations, unlike other HA configurations that allow only one secondary server. The model pages, however, say to keep only one read-only server in the shared-database setups, so check the page for the model you pick before planning several.

Which PostgreSQL guide matches your build

The PostgreSQL guides split by build number:

  • Builds 8400 and above: the guides for the Secondary Server model with Read-Only Server continuity. These use PostgreSQL native streaming replication. ManageEngine states that the earlier Rubyrep-based replication model has reached end-of-life.
  • Builds 8310 and below: a separate, older page. The monitoring page for the older Secondary Server model says it applies to builds 8310 and below.

If you already run the older "Secondary Server Model using PostgreSQL Database" and want the newer design, the newer page lists a prerequisite: remove the existing configuration first, using the linked removal steps. Take a full backup of the PAM360 application and database before you start.

The three models for builds 8400 and above

The newer PostgreSQL page points to three model pages:

  1. Primary and Secondary Server with a Shared Database and an Additional Read-Only Server. ManageEngine marks this one as recommended. It combines the failover of an active pair with a read-only copy for disaster recovery.
  2. Primary Server with a Read-Only Server. One server handles all writes. The read-only server is a replica that gets updates in near real time and can be promoted if the primary is lost. It can also take reporting reads.
  3. Primary and Secondary Server with a Shared Database. Both servers are read-write and connect to one shared database, so either can carry on if the other fails.

The failover behavior described for the combined design is easier to see as a table:

SituationWhere users connectAccess
Primary fully activePrimaryRead and write
PAM360 service down on the primary, database still upSecondaryRead and write
Primary completely downRead-only serverRead-only
Primary and secondary both downRead-only serverRead-only

Two further points from that page matter for planning. Scheduled tasks run only on the primary server. And in the shared-database model without a read-only server, an outage of the primary makes stored passwords unavailable, because the database runs on the primary.

Prerequisites and ports

For the PostgreSQL models on builds 8400 and above, ManageEngine lists these prerequisites:

  • Remove any existing "Secondary Server Model using PostgreSQL Database" configuration.
  • Open the web server port, 8282, on all servers.
  • Open the database port, 3456, on the primary and read-only servers. If you use custom ports, open those instead.
  • If Application Gateways are deployed, install the Microsoft Visual C++ Redistributable for Visual Studio 2015 or later on the secondary or read-only server.
  • Keep only one read-only server; remove any others.
  • Store pam360_key.key in a path every server can reach, if possible.

The PAM360 architecture page gives the wider port table. The entries relevant to an HA design are:

PurposePortDirection
Web client and REST API8282 TCPUser machine to PAM360 server
Bundled PostgreSQL, replication between primary and read-only servers3456 TCPPAM360 server to PAM360 server
Application Gateway, password management8288 TCPGateway server to PAM360 server
Application Gateway, remote sessions8289 TCPGateway server to PAM360 server
Session Gateway (web and mobile sessions)8283 TCPUser machine to PAM360 server
HTTPS gateway to web resources8285 TCPUser machine to PAM360 server
SSH API6622 TCPUser machine to PAM360 server

Those are defaults. If you changed any of them, use your values. The architecture page also says replication between the primary and read-only servers uses PostgreSQL port 3456, and it names 8288 and 8289 for the Application Gateway deployment model used across isolated networks or DMZ regions.

Port 5432 appears in the same table for a different purpose: PAM360 connecting to a PostgreSQL server it manages (password rotation), not replication between PAM360 servers. Do not open it between PAM360 servers on the strength of that line.

The Microsoft SQL Server guide lists no firewall ports. It mentions 1433 as the default when you give no instance name or custom port, and it says you can append :<port> to a host name for a dedicated instance.

Set up the PostgreSQL shared-database pair with a read-only server

The steps below paraphrase ManageEngine's page for the recommended model. Run them in this order, because the setup packs are generated on the primary and consumed by the other servers.

Part 1: the primary server

  1. Install PAM360 and select the High Availability Primary Server configuration.
  2. Start the PAM360 service so the server and database initialize.
  3. If PAM360 already runs, disable any existing HA configuration and keep at most one read-only server. If the key file is in a remote directory, set its full path in manage_key.conf.
  4. Stop the PAM360 service. In the bin folder, run HAPostgreSQLAppSetup as an administrator (.bat on Windows, .sh on Linux), passing the FQDN or IP address of the secondary server.
  5. The script runs the database as a standalone service named PAM360 PostgreSQL and creates AppServerPack_<FQDN/IPAddress>.zip in the replication folder.
  6. Still in bin, run ROSetup and provide the read-only server's IP address, user name, password and a unique slot name. This creates ROPack_<Slotname>.zip in the replication folder.
  7. Import three certificates with importCert: ServerCer.cer and CAcert.pem from conf, and ServerCer.cer from agent. Change the paths if you use custom certificates.
  8. Restart the PAM360 PostgreSQL service, then start the PAM360 service.

The standalone database is started and stopped with start_pgsql and stop_pgsql in the bin folder.

Part 2: the secondary server

  1. Install PAM360 and select High Availability Primary Server. The FQDN or IP address must match the one you gave when generating the pack. Do not start the service.
  2. Copy AppServerPack_<FQDN/IPAddress>.zip to the installation directory and extract it. ManageEngine notes this overwrites files with the primary's configuration.
  3. Point manage_key.conf at the key file. If there is no shared path, copy the primary's key file to a local directory and set that path.
  4. Start the PAM360 service.

Part 3: the read-only server

  1. Install PAM360 and select Read-Only Server.
  2. Extract ROPack_<Slotname>.zip into the installation directory.
  3. Set the key file path in conf/manage_key.conf, creating the file if it is missing.
  4. Import the same three certificates with importCert. If you use a custom SSL certificate, copy it from the primary into the read-only server's conf folder.
  5. Start the PAM360 service.

Verify

Sign in to the primary server and open Admin >> Business Continuity >> High Availability to see the status of all configured servers. For a primary with a read-only server only, the page gives Admin >> Business Continuity >> Read-Only Server as the monitoring location.

Primary  (read/write)  --- shared database --- Secondary (read/write)
   |                                              
   +--- streaming replication, port 3456 ---> Read-Only Server (read only, promotable)

Set up the Microsoft SQL Server secondary server

The SQL Server guide uses a different tool chain: a replication wizard on the primary and a database switch on the secondary.

Prerequisites

  • The secondary SQL Server database must use SSL, because PAM360 requires SSL for communication between the databases.
  • The primary and secondary database logins must hold the sysadmin role. Windows credentials can be used from Windows.
  • A dedicated named instance on a specific port is recommended, since PAM360 connects only in SSL mode.
  • On build 8000 and above the master key is in pam360_key.key; on earlier builds it is in masterkey.key in the conf folder. On build 8000 and above, keep the key at the path set in manage_key.conf. ManageEngine warns the service may not start without it.

Configure replication on the primary

  1. Stop the PAM360 primary server.
  2. Import the secondary database's SSL certificate with importCert, passing the absolute path of the certificate.
  3. Copy the SQL master key from the key file.
  4. In bin, run ConfigureReplication.bat (or sh ConfigureReplication.sh on Linux) to open the wizard.
  5. Under Master Details enter the master host name, instance name, a sysadmin login name and password, the master database name and the master key. Under Slave Details enter the matching values for the secondary. The secondary database name defaults to pam360standby.
  6. Click Test, then Configure. ManageEngine says this can take 30 minutes or more.
  7. Restart the primary server.

Configure the secondary

  1. Install PAM360 and choose to configure the server as a High Availability Secondary Server. Do not start the service.
  2. Run ChangeDB.bat (or sh ChangeDB.sh) from the secondary bin folder. Set Server Type to SQL Server and enter the host name, instance name and the exact database name used during replication setup. ManageEngine recommends Windows authentication because it stores no credentials; with SQL authentication the credentials are saved in database_params.conf.
  3. Copy SQLServerHAPack.zip from the primary and extract it into the secondary installation directory. This carries the license, custom icons and rebranding.
  4. Edit manage_key.conf in the secondary conf folder to point to pam360_key.key.
  5. Start the secondary server.

To verify, open Admin > General > Secondary Server on either server. "High Availability Status: Alive" confirms the setup.

Monitoring and alerts

The monitoring page for the older Secondary Server model shows an HA summary, up or down icons for each server and database, and a replication pending count. A pending count of zero means the servers are in sync. When the HA state is inactive, usually because of a network problem, replication stops and then resumes on its own once the connection returns; ManageEngine says users on either server are not affected during the disruption.

For alerts, the page points to Audit >> Resource Audit >> Configure User Audit >> General Operations. For the events "High Availability Alive" and "High Availability Failed" you can choose email, SNMP trap or Syslog. That page is written for builds 8310 and below, so confirm the same path exists in your build before you rely on it.

Two cautions from the same page: changing the primary server's port breaks HA until you reconfigure the secondary with the new port, and enabling or changing the two-factor authentication type in an HA setup requires restarting the secondary server.

Troubleshooting and planning checks

ManageEngine's pages do not publish a list of error messages for these setups, so use these checks drawn from the documented requirements:

  • Secondary will not start after extracting the pack. Confirm the FQDN or IP used for the pack matches the secondary, and that manage_key.conf points to a reachable pam360_key.key.
  • Read-only server cannot replicate. Check that port 3456 is open to the primary and read-only servers, the replication slot name is unique, and the certificates were imported on both.
  • Read-only URL does not redirect. ManageEngine warns that redirection from the read-only server to the primary or secondary may fail across networks and recommends using the primary URL in that case.
  • After a failure, before restarting recovered servers. The page says to restore the database backup before restarting them, to avoid data loss. Promoting a read-only server to primary is a manual procedure on the page, involving stopping the service, removing standby.signal and editing configuration files, so rehearse it away from production.

For a comparison with another vendor's failover design, see the EpiFive guide on Azure SQL failover groups.

Checklist

  • Pick the model: shared database pair plus read-only server is the recommended PostgreSQL design; SQL Server uses the secondary server guide.
  • Match the guide to your build: 8400 and above, or 8310 and below.
  • Remove any older Rubyrep-based HA configuration first, after a full backup.
  • Open 8282 on all servers and 3456 on the primary and read-only servers, or your custom ports.
  • Keep pam360_key.key reachable and manage_key.conf correct on every server.
  • Verify in the console, then test failover on a non-production copy before you depend on it.

What we checked

  • Opened ManageEngine's "High Availability and Disaster Recovery Models in PAM360" page on 2026-11-20 and copied the four model names; the database column in our table comes from the linked guide titles, as the page text does not state it outright.
  • Opened the PostgreSQL read-only server continuity page on 2026-11-20 and confirmed the build 8400 and above scope, the Rubyrep end-of-life statement, and ports 8282 and 3456.
  • Cross-checked port 3456 between the PostgreSQL guide and the PAM360 architecture page: both give it for replication between the primary and read-only servers.
  • Compared the setup steps on the shared-database-plus-read-only page and the Primary-with-Read-Only page; both use the ROSetup script and ROPack_<Slotname>.zip.
  • Opened the Microsoft SQL Server secondary server page and noted the verification text "High Availability Status: Alive" and the 30-minute-or-more replication note.
  • Opened the monitoring page and noted it is scoped to builds 8310 and below; we flagged that in the article rather than assuming it applies to newer builds.
  • Not tested on a live PAM360 installation; steps follow ManageEngine's documentation as of 2026-11-20, and the pages do not state licensing requirements, so none are given here.

Last tested: 2026-11-20

References

Questions people ask

What ports does PAM360 high availability need?

ManageEngine lists 8282 (web client) and 3456 (bundled PostgreSQL, used for replication between the primary and read-only servers) as the defaults to open between servers, and says to use your custom ports if you changed them. If you deploy Application Gateways, the architecture page also lists 8288 and 8289. The SQL Server guide lists no firewall ports; it only refers to port 1433 or your custom instance port.

What is the difference between a secondary server and a read-only server in PAM360?

In the shared-database model, the primary and secondary servers are both read-write and connect to one database. A read-only server is a replica fed by PostgreSQL streaming replication; it serves read-only access and can be promoted if the primary is lost. The documented recommended design combines both.

How do I check that PAM360 HA is working?

For the PostgreSQL models, sign in to the primary server and open Admin >> Business Continuity >> High Availability. For the Microsoft SQL Server secondary server model, open Admin > General > Secondary Server on either server and look for "High Availability Status: Alive".

Can PAM360 high availability use the bundled PostgreSQL database and Microsoft SQL Server?

Yes, ManageEngine documents separate guides for both. The newer read-only server models are documented for PostgreSQL on builds 8400 and above, while the Microsoft SQL Server guide uses a secondary server with database replication configured through a ConfigureReplication wizard.

ManageEngine PAM360High AvailabilityPostgreSQLMicrosoft SQL ServerPrivileged Access Management