So what happens here is the upgrader worker watches the model config for changes. If the version changes, it simply requests tools for the version.
Nowhere is it stored that we requested the upgrade from the specific stream, instead at this point, we just use the configured model value (default=released) for "agent-stream".
Changing this value to mirror that supplied at the CLI allows the upgrade to proceed.
So what happens here is the upgrader worker watches the model config for changes. If the version changes, it simply requests tools for the version.
Nowhere is it stored that we requested the upgrade from the specific stream, instead at this point, we just use the configured model value (default=released) for "agent-stream".
Changing this value to mirror that supplied at the CLI allows the upgrade to proceed.