tickets/SP-2785: Get the schedview snapshot viewer working again

[ehneilsen]

Two changes were required:

1. I updated lsst-sqre/build-and-push-to-ghcr to the latest tagged version. This got the container to build, but it still couldn't load snapshots.
2. Remove the profile specification from the localized snapshot URIs, because the way the container app is set up it needs to be the default. But, running on a USDF devel node still needs to be set to use the lfa profile. The profile the dashboard uses can still be set with the AWS_PROFILE environment variable, so if it fails and it looks like this was the issue, there's a longer error message suggesting setting it.

[rhiannonlynne]

Just to clarify ..

Remove the profile specification from the localized snapshot URIs, because the way the container app is set up it needs to be the default.

is because the secret being injected from vault is the "default" AWS credentials, and not the "lfa" credentials (but default works ok for reading)?
I'm less clear on why the default doesn't work from the dev nodes.

[ehneilsen]

Just to clarify ..

Remove the profile specification from the localized snapshot URIs, because the way the container app is set up it needs to be the default.

is because the secret being injected from vault is the "default" AWS credentials, and not the "lfa" credentials (but default works ok for reading)? I'm less clear on why the default doesn't work from the dev nodes.

Yes, it needs to use the default profile in the credential file provided to the phalanx container. This is not the same as the default profile on the USDF devel nodes, but a special purpose credential file created just for this app. See the discussion in the relevant pull request.
I'm pretty sure the default profile in the USDF devel nodes also used to work, but it seems not to work anymore: we need to use the lfa profile in the devel nodes instead.