Problems during install.sh run

[madsklitgaard]

I'm trying to install Tator using the install guide on tator.io

Running the install.sh script and everything seems to run fine until it gets to

client.go:250: [debug] Created a new CronJob called "expire-invitations-cron" in default
client.go:250: [debug] Created a new CronJob called "expire-announcements-cron" in default
client.go:250: [debug] Created a new CronJob called "expire-password-resets-cron" in default

Then I just get this and in the end it just stalls

wait.go:66: [debug] beginning wait for 63 resources with timeout of 1h0m0s
ready.go:258: [debug] Service does not have load balancer ingress IP address: default/nginx-svc
ready.go:258: [debug] Service does not have load balancer ingress IP address: default/nginx-svc
ready.go:258: [debug] Service does not have load balancer ingress IP address: default/nginx-svc
....

Could anybody help me? I've tried setting the IP address using the env HOST_IP but that doesn't help.

I'm using a clean VPS with Ubuntu 20.04 - 4 cores and 16GB Ram

[jrtcppv]

Hi there, we actually will be switching this repository to a Docker Compose based deployment for exactly this reason; many users report issues related to Kubernetes and network configuration. Although it is not yet in an official release, the compose deployment has been pushed to the main branch. Would you be interested in trying that? The steps would be to check out the main branch, then follow the steps outlined in the quick start:

git clone --recurse-submodules https://github.com/cvisionai/tator
cd tator
make tator
make superuser

In addition, we have written preliminary documentation for the compose deployment, which you can access in deploy previews:

• Installing Tator
• Using a domain and certificate
• Migrating from Kube to Compose

Let me know if you want to give that a try, I think that would be a better path than debugging Kube's interactions with your network setup.

[jrtcppv]

Also I wanted to add, if you prefer to stay on a stable release I'd be happy to help you work through the install script issues in 0.2.23. Just let me know which path you prefer.

[madsklitgaard]

Thank you very much for the reply - I'll just try to run the Docker based version that you refer to above first and see if I can't get that to work instead :)

[jrtcppv]

Sounds good, let me know how it goes.

[madsklitgaard]

The install works perfectly now. It's up and running on a custom domain. Only thing now is I'm having some trouble uploading video files to project without receiving a "Failed to fetch" error. But I'm guessing I'll need to look into change some limits somewhere.