In my CI, ~half of the tests are failing with
error: requested a non-existing branch on 3.6/edge for snap "juju": rpcerrors
And the other half are passing with
juju (3.6/edge/rpcerrors) 3.6.15 from Canonical** installed
How can this be?
Example:
AFAICT the store returned revision-not-found. Then depending on the additional context attached to the error snapd tries to produce more useful message.
I would try to figure out why the store claims there 3.6/edge/rpcerrors does not exist. If it’s during installation, then snapd pokes this Refresh endpoint with action = “install”. You could try issuing the request yourself and to see whether the results are consistent. Another option is adding SNAP_HTTP_DEBUG=7 to snapd’s env to see the content of the requests/responses, maybe there’s some additional context the store provides.
In any case, there’s not much more we can do in snapd.
Thank you @mborzecki1 . I wasn’t able to reproduce this locally, but also I haven’t tried “hard enough” (for example creating a loop that installs and uninstalls until the error appears).
Should I move this to the store category? Or repost it there?
I changed the topic category. I you can, I would recommend reaching out to store folks directly. AFAIU they can access the logs and should be able to figure out what’s going on here.