Requirements

The Server Host is the machine where the Job Server process (jobserverd) runs. We recommend selecting a dedicated physical or virtual machine with restricted access to users.

The following requirements should be met for the Server Host:

Supported OS

The Server Host must be run using a Linux operating system. See the System Requirements for more information on supported operating systems.

Queuing System

The jobserverd process must be able to submit batch jobs to the queuing system.

Confirm that the executable files of the batch queuing system (e.g. qsub/sbatch, qdel/scancel, etc.) are available on the Server Host and the directory containing these executables is contained in the PATH environment variable.

Access to the Schrödinger Suite Installation

Confirm that the Schrödinger Suite installation directory is available on the Server Host and executables contained in the Schrödinger Suite installation can be executed.

Note: Job Server release 2026-1 can support jobs submitted from applications of the past five Schrödinger Suite release versions.

Network Connections

The Server Host requires two open ports for communicating with the Submission Host and Execution Hosts:

• the Job Server API port (default: 8030/tcp)
• the I/O port (default: 8029/tcp)

In addition, the Job Server needs to be able to connect to the following services on these ports:

• PostgreSQL database server (default port: 5432/tcp)
• LDAP(S) server (default port: 389/tcp for unencrypted LDAP, port 636/tcp for encrypted LDAPS).

Authentication

Users must be able to authenticate with at least one of the following methods:

• LDAP: the Job Server authenticates the user with an LDAP server. Users do not need to have interactive login access to the Server Host on which the Job Server process is running.
• Unix sockets: the Job Server authenticates the user via sockets. Users need to be able to log in to the Server Hosts interactively (typically with SSH). This is a one-time requirement for authentication only. Users do not need interactive login access to the Server Host to submit jobs.

Job database

Note: The job database should be configured on a local disk. If it isn't, then you will likely have performance issues when you try to run larger numbers of jobs.

For large HPC deployments (concurrent use of more than 1000 GPUs or CPUs), you will need to do this additional step: Configuring PostgreSQL as the Job Database

If you are installing Job Server on a single-node machine, use SQLite instead.

Disk space for file store

Note: The file store should be configured on a local disk. If it isn't, then you will likely have performance issues when you try to run larger numbers of jobs.

Selecting a fast, local disk as file store for the Job Server increases performance. You need enough space to hold all your completed jobs until they are downloaded. Please contact Schrodinger if you need advice on this.

We strongly recommend the ability to dynamically increase the size of file store for future expansion.

Job Server User Account

Create a dedicated job server user account for administration use. We will refer to this "jobserver" in the following sections. This user needs to be able to run the submit, cancel, and status commands for your queueing system.

Directory where the Job Server files should be installed

We refer to this as the <jobserver_dir> in the following sections

Process management

systemd is used for automated service administration

Root access to install