OHDSI Home | Forums | Wiki | Github

Some advice needed re installing Atlas on Linux VM

Implementers
#1

Hello,

I need some advice. I am currently trying to set up the infrastructure to perform QC on our OMOP database. I have successfully installed and ran DQD, Achilles, and CdmInspection, and now want to install Atlas. As far as I can gather there are several ways to go about this, and am not sure what is the best option for me. I am working on a Linux VM with Ubuntu 18.04. A complicating factor is that this VM is in a secure environment that has only very few IP addresses whitelisted. I can pull for example pull R packages from CRAN, but have no access to GitHub. What would be the best way for me to setup Atlas? Any suggestions are appreciated!

Cheers,
Bert

#2

Within a VM that has internet access you could use Docker… One docker container for Atlas’ configuration db, another for Tomcat 8.5, another docker container to compile node, another to compile with mvn

#3

Are you sure the whitelist is working at the IP address level? Can you access Github repositories on a web browser? Is it curl/wget that isn’t working? Are you using the ‘remotes’ library, if so, have you generated and set a github_pat as described here?

I ask because seems odd that CRAN would be whitelisted at all. My guess is the administrator of the environment actually has set up their own repository for R packages. However, if that isn’t the case, often the problem is SSL certs for curl/wget/remotes.

If it is the case this is a whitelist/IP limitation, here are a few alternative routes:

  1. In some secure environments, there are prescribed upload portals. Depending on policy, you can upload approved software and packages through that upload portal. You can then install the packages from source.

  2. Many IT Depts. do not like Docker for security reasons. The Broadsea team recently created a Podman deployment option as an alternative. Podman does not require root privledges like Docker does.

  3. You can ask your IT Dept. to provision a VM following the OHDSI in a box or OHDSI in AWS architecture.

You should always check with your Security Officer / IT Dept for options and assistance and approval. Even if you find a workaround it may be against policy or worse expose sensitive information.

t