Hello, I solved previous problems that happened when I first tried installing AODT.
I thought everything was fine, but unfortunately when I followed the docs tutorial step by step(opening tokyo.usd and creating panels and RU/UE), I noticed that the clickhouse didn’t work.
The connection failed continuously, so I decided to reinstall from the beginning.
(with the “./aodt_bundle/install.sh localhost $NGC_CLI_API_KEY” command)
After running the installation again, the aodt launched and the clickhouse connection succeeded.
However, while trying the tutorials again, this time the Ray Properties selection disappeared ^^;;
Also, this time this error happens when I click the tool bar.(“There is no worker available or connection couldn’t be established.”)
I noticed one bad news from the terminal shown as below.
I checked out other similar situations and solutions given, and checked that the clickhouse exists.
There weren’t any similar cases about “no worker” problems neither “Ray properties” disappearance.
Please help me solve this situation.
@jpeace321 What was the previous problem and resolution?
This is not a clickhouse error. As you rightly concluded, the clickhouse is running. It is the backend container that has failed.
Can you please share:
- nvidia-smi
- Backend console log when you bring up the backend docker process.
The previous problem was about loading the zip file. I just solved it by getting a GUI and downloading it online with chrome.
-
nvidia-smi
-
I’m not sure which console log you need to check, so I captured three possibly useful consoles.
Console after launching AODT.
Console log that appeared right after the reinstallation.
Error log that I found that the docker “backend_bundle-clickhouse-1” wrote.
If none of the consoles are helpful, tell me what to do for more information!
Thank you.
Hi @jpeace321
did the installation finish successfully when you tried it for the second time?
it appears that the second GPU is not detected by the system, hence seeing the issue connecting to the backend.
What GPUs you have on your setup and what’s your systems configuration?
Sorry for the late response.
The second installation finished with the “Error response from daemon”. (Above screenshot)
I guess that’s why the worker wasn’t available.
My configurations are as below
What steps should I go through to solve this problem?
@jpeace321 I dont see the screen shot for “error response from demon”. Can you please post it?
Additionally, please post the output of running ‘docker ps’ command on backend.
The above image I mentioned was the previous comment I answered. (coppied it below)
The result of “docker ps” is like this.
Thank you for the fast response.
Hi @jpeace321
The backend container is not running form the above screenshot.
Can you please provide the output for “nvidia-smi”?
This is the screenshot of “nvidia-smi”.
Hi @jpeace321
Is there only 1 GPU available on your setup or you have 2 GPUs for running backend and frontend simultaneously?
@jpeace321 If there is only a single GPU then you need to run in single GPU mode. That means that the front end and backend cannot be running simultaneously. Do the following:
- Make sure there is no backend running
- Open the UI, set up the stage and generate the UEs. This will prepare the database for running in the db_replay mode
- Close the UI. Start the backend
- Run the simulation in the db_replay mode. See documentation for " Database Replay"
1 Like
