ATLAS PanDA cache instances

Operating systems and python versions

PanDA cache is currently running on CentOS7 instances and python3.10.


The nodes are aipanda047 and aipanda048.


The cache is a skimmed PanDA server running just a subset of functionalities. The full installation guide is the same as PanDA server: The objective of this wiki is just to give an overview of the ATLAS production instances.

Python virtual environment

System env variable: VIRTUAL_ENV

Location: /opt/panda

/opt/panda/bin/python -V
Python 3.10.7

The PanDA code will be under: /opt/panda/lib/python3.10/site-packages/pandaserver

Installing and updating the code

The first time you install the code, you will want to bring in all atlasprod dependencies to install e.g. cx_oracle and the Rucio client:

/opt/panda/bin/pip install panda-server[atlasprod]

In order to install just the latest PanDA server code from github:

/opt/panda/bin/pip install --no-deps --force-reinstall git+

systemd service

The environment for systemd services has to be passed in a new format. The environment file is /etc/sysconfig/panda_server_env.

PanDA cache only runs the httpd service (not the daemons). The service unit file for ATLAS is /etc/systemd/system/panda.service

The very first time after setting up a machine, you need to enable the services.

systemctl enable panda.service

You can start/stop/restart the service like this:

systemctl start panda.service
systemctl stop panda.service
systemctl restart panda.service

Systemd will not print anything out to the console. Instead you need to query the output by running:

systemctl status panda.service

Here you will find information, for example if the DB Schema check was passed.

If there are issues starting the service, you can get additional information using journalctl.

journalctl -xeu panda.service