WIP: Clean /var after Anaconda #107
Conversation
This way we are truly starting from a clean slate. It means we know we won't have a `random-seed` for example. OSTree is designed to support this but unless we use it *by default* we aren't really testing it. This actually now breaks things since currently the creation of the `core` user is in `image.ks`, which we need to fix: coreos/afterburn#90 (comment)
| shift | ||
|
|
||
| coreos_gf_run_mount "${src}" | ||
| coreos_gf glob rm-rf ${stateroot}'/var/*' |
There was a problem hiding this comment.
do we need a coreos_gf_shutdown like in gf-oemid ?
|
if we make this change does it make it harder to use the assembler for el7 ? IOW can el7 handle the start with empty /var/ as good as Fedora can? |
Maybe, let's not concern ourselves too much with that quite yet. I mean, let's consider it but we can't block on it. If we need to carry hacks we'll figure out whether to add them here later, or fork this repo for el7, etc. |
that's fine. 👍 you added the "help wanted" label. Can you describe what you're looking for? |
Change our build or config process to have Ignition create the |
|
I forgot about this PR and ended up reimplementing it in #205 |
This way we are truly starting from a clean slate. It means
we know we won't have a
random-seedfor example. OSTree isdesigned to support this but unless we use it by default
we aren't really testing it.
This actually now breaks things since currently the creation of
the
coreuser is inimage.ks, which we need to fix:coreos/afterburn#90 (comment)