-
Notifications
You must be signed in to change notification settings - Fork 28
snapd run from refresh-bits panics in InternalToolPath #25
Comments
Based on a conversation just a moment ago this is a self-defense mechanism that has gone a bit to far. As for devtools, I haven't used them myself in a while so there are probably a whole lot of things gone stale. For smaller things that I usually work with I resort to |
This is a bit of a hack, but the following seems to work to temporarily test out a new build of snapd without overwriting the distro packaged version:
And to reverse the process:
|
I also stop the socket |
I used them successfully by setting the CMD_PREFIX to an empty string. I am using ubuntu-core inside a virtualbox vm for testing, so I am not afraid to rollback to a snapshot. But in general I didn't have problems with the automatic devtools rollback: Line 45 in 02c3f00
The problematic line in the snapd is: https://github.com/snapcore/snapd/blob/master/cmd/cmd.go#L128 If I interpret it correctly, it would be enough to rewrite the CMD_PREFIX as a "CMD_SUFFIX" (code has to be changed accordingly) with the value ".test". |
I was trying to use refresh-bits to test out a change to snapd, and had snapd panic with the following message:
This happened when I was trying to connect an interface that changed the mount configuration for the snap. The call in question seemed to come from
interfaces/mount/ns.go
, which doesn't seem to like the way refresh-bits builds snapd.The text was updated successfully, but these errors were encountered: