RapidMiner AI Hub Error: RMSSOAuthenticationError

rfuentealba
rfuentealba New Altair Community Member
edited November 5 in Community Q&A
Hello all,

Long time no see! I was testing the latest version of RapidMiner AI Hub on Docker, when I found this error message:



All I did was following the instructions:

  • Changed the port to 8080.
  • Changed the proxy port to 8080 on the docker compose file.
  • Changed the port to 8080:80 on the docker compose file too.
  • Added my server license.
  • Added my go license.
No avail. I have been reviewing this for a couple hours and I am clueless about it. I don't even know what else to try or what kind of information to provide.

This is a Microsoft Windows 11 with WSL2 enabled, 64 Gb RAM, i9 processor, so the issue is a little over the top of my head. I'm going to try the same on a Mac in case it's a thing with Windows, but I'm not sure about that.

Anyone can help me? Thanks in advance.

Tagged:

Best Answer

  • rfuentealba
    rfuentealba New Altair Community Member
    Answer ✓
    Hello Everyone,

    Good news! I was able to make it run on Windows!

    What I did is pretty much what the documentation says:

    On the docker-compose.yml file, 
    • Went to the rm-proxy-svc service definition, and switched the port from 8080 to 2080.
    • Configured the PROXY_HTTP_PORT section on the rm-server-svc service definition to 2080, not to 8080. Why? I don't know, but when I set the port to 8080, it crashed everything and I don't know why.
    On the .env file,
    • Modified the PUBLIC_DOMAIN to my IP address. It allows hostnames too, as long as Docker can reach these.
    • Modified the PUBLIC_URL and SSO_PUBLIC_URL to http://my.ip.address:2080/. This one was suggested by @MarcoBarradas
    • Modified the COMPOSE_PROFILES to proxy,landing-page,jupyter,grafana,rts,platform-admin,server,job-agent,ces,token-tool. I left go, tabgo and radoop out. This one was suggested by @MarcoBarradas too.
    • Copied the content of the echo $RANDOM | md5sum | head -c 20; echo | base64; command I executed on WSL Ubuntu to the AUTH_SECRET variable.
    • Marco also suggested to complete all the SECRETs, but I didn't find this necessary, as the Docker compose does that for you.
    I will keep testing this. If you run RapidMiner AI Hub on Docker Desktop for Windows (I haven't tried UNIX yet), don't use any of the ports defined in the .yml, as Docker just doesn't like the repeated port.

    To run it, exactly as @BalazsBarany said:
    • docker-compose up -d rm-init-svc
    • docker-compose logs -f rm-init-svc
    There will be a line saying [RapidMiner Server initialization] Waiting for RapidMiner Server startup and license load to initialize python configuration. Only when this line appears, execute the command:
    • docker-compose up -d
    Leave RapidMiner running for some 10-15 minutes. It doesn't start right away because it has to coordinate a couple things.

    Another thing I forgot:

    If you are running this on Windows 10/11, create a file named .wslconfig and put this:

    [wsl2]
    memory=32GB
    processors=4
    localhostforwarding=false

    You will need quite a lot of memory to run this, and Docker Desktop comes handicapped on Windows 10/11, so you'll have to rely on the WSL kernel to run it. This is the way to assign more memory. Beware that if you don't have 32 Gb of RAM to spare, your computer will use the swapfile heavily.

    All the best,

    Rod.

Answers

  • rfuentealba
    rfuentealba New Altair Community Member
    Update:

    I did the same process on a macOS machine, and it shows the exact same error. Retried following everything carefully and no, it's not Docker. I added more memory (32 Gb) to the Docker Desktop on both my PC and my Mac and nope. Still getting the RMSSO error.

    :(
  • BalazsBaranyRM
    BalazsBaranyRM New Altair Community Member
    Hi Rodrigo!

    I usually see this when the Docker deployment is not done yet. Starting up everything takes a few minutes even on a powerful server. But it works after a few minutes. 

    This should be better documented and the proxy component should probably have better error reporting.

    Regards,
    Balázs
  • rfuentealba
    rfuentealba New Altair Community Member
    Hi Balazs,

    I can confirm this is not the case. I have reviewed the documentation entirely, even saw a video where you make the RapidMiner AI Hub container and I do the exact same things, to no avail.

    I'm still searching just in case something else comes to light, still without luck.

    Thank you!
  • Marco_Barradas
    Marco_Barradas
    Altair Employee
    Hi @rfuentealba
     
    Please check that you have set all the SECRETS on the .env file and that you have changed the port in the two places it should be done on the .yml file.



  • rfuentealba
    rfuentealba New Altair Community Member
    Hello @MarcoBarradas,

    Apparently, my issue isn't any of that. I've spoken with Anthony, and he told me Windows isn't supported to run RapidMiner AI Hub on Windows. There have been people able to run it, but after a slight change everything went down the drain again. I'll keep figuring out if I can fix this though.

    All the best,

    Rod.
  • rfuentealba
    rfuentealba New Altair Community Member
    Answer ✓
    Hello Everyone,

    Good news! I was able to make it run on Windows!

    What I did is pretty much what the documentation says:

    On the docker-compose.yml file, 
    • Went to the rm-proxy-svc service definition, and switched the port from 8080 to 2080.
    • Configured the PROXY_HTTP_PORT section on the rm-server-svc service definition to 2080, not to 8080. Why? I don't know, but when I set the port to 8080, it crashed everything and I don't know why.
    On the .env file,
    • Modified the PUBLIC_DOMAIN to my IP address. It allows hostnames too, as long as Docker can reach these.
    • Modified the PUBLIC_URL and SSO_PUBLIC_URL to http://my.ip.address:2080/. This one was suggested by @MarcoBarradas
    • Modified the COMPOSE_PROFILES to proxy,landing-page,jupyter,grafana,rts,platform-admin,server,job-agent,ces,token-tool. I left go, tabgo and radoop out. This one was suggested by @MarcoBarradas too.
    • Copied the content of the echo $RANDOM | md5sum | head -c 20; echo | base64; command I executed on WSL Ubuntu to the AUTH_SECRET variable.
    • Marco also suggested to complete all the SECRETs, but I didn't find this necessary, as the Docker compose does that for you.
    I will keep testing this. If you run RapidMiner AI Hub on Docker Desktop for Windows (I haven't tried UNIX yet), don't use any of the ports defined in the .yml, as Docker just doesn't like the repeated port.

    To run it, exactly as @BalazsBarany said:
    • docker-compose up -d rm-init-svc
    • docker-compose logs -f rm-init-svc
    There will be a line saying [RapidMiner Server initialization] Waiting for RapidMiner Server startup and license load to initialize python configuration. Only when this line appears, execute the command:
    • docker-compose up -d
    Leave RapidMiner running for some 10-15 minutes. It doesn't start right away because it has to coordinate a couple things.

    Another thing I forgot:

    If you are running this on Windows 10/11, create a file named .wslconfig and put this:

    [wsl2]
    memory=32GB
    processors=4
    localhostforwarding=false

    You will need quite a lot of memory to run this, and Docker Desktop comes handicapped on Windows 10/11, so you'll have to rely on the WSL kernel to run it. This is the way to assign more memory. Beware that if you don't have 32 Gb of RAM to spare, your computer will use the swapfile heavily.

    All the best,

    Rod.