Coder ssh during provision results in error after wait: workspace "test" has no agents #5836
Description
At first we stream the build log, which is fine. But after it has finished, there is a short interval during which workspace agents don't yet exist. The SSH command doesn't like this.
Ultimately we should probably not consider the provisioning done until the database is fully prepared.
> coder ssh test
✔ Queued [255ms]
✔ Setting up [658ms]
⧗ Planning infrastructure
data.coder_provisioner.me: Refreshing...
data.coder_workspace.me: Refreshing...
data.coder_provisioner.me: Refresh complete after 0s [id=33a5d798-7c27-4e59-b847-0f2fcc5eb291]
data.coder_workspace.me: Refresh complete after 0s [id=62d19979-744d-4651-9ac6-9ce8352c0859]
coder_agent.main: Refreshing state... [id=39b456e6-7317-4b02-bd6a-100d4fe86e65]
coder_agent.main: Refresh complete [id=39b456e6-7317-4b02-bd6a-100d4fe86e65]
coder_agent.main: Drift detected (update)
null_resource.workspace[0]: Plan to create
Plan: 1 to add, 0 to change, 0 to destroy.
✔ Planning infrastructure [1401ms]
⧗ Starting workspace
Terraform 1.3.4
null_resource.workspace[0]: Plan to create
null_resource.workspace[0]: Creating...
null_resource.workspace[0]: Provisioning with 'local-exec'...
null_resource.workspace[0]: (local-exec): (output suppressed due to sensitive value in config)
null_resource.workspace[0]: (local-exec) Provisioning complete
null_resource.workspace[0]: Creation complete after 5s [id=285218828837375588]
Apply complete! Resources: 1 added, 0 changed, 0 destroyed.
Outputs: 0
✔ Starting workspace [6158ms]
✔ Cleaning Up [108ms]
workspace "test" has no agents
Run 'coder ssh --help' for usage.