Rootless rpm-ostree can't mknod; causes broken RHCOS #1950
Assignees
Comments
cgwalters
added a commit
to cgwalters/rpm-ostree
that referenced
this issue
Dec 10, 2019
Rather than giving dracut `cap_mknod` which won't work in unprivileged scenarios, append a tiny static pre-generated CPIO blob with `/dev/random` and `/dev/urandom` to the output of dracut. This is a hack until dracut does this itself. But the problem is patches to dracut will take eleven billion years to ship in RHCOS. Closes: coreos#1950
cgwalters
added a commit
to cgwalters/rpm-ostree
that referenced
this issue
Dec 10, 2019
Rather than giving dracut `cap_mknod` which won't work in unprivileged scenarios, append a tiny static pre-generated CPIO blob with `/dev/random` and `/dev/urandom` to the output of dracut. This is a hack until dracut does this itself. But the problem is patches to dracut will take eleven billion years to ship in RHCOS. Closes: coreos#1950
openshift-merge-robot
pushed a commit
that referenced
this issue
Dec 10, 2019
Rather than giving dracut `cap_mknod` which won't work in unprivileged scenarios, append a tiny static pre-generated CPIO blob with `/dev/random` and `/dev/urandom` to the output of dracut. This is a hack until dracut does this itself. But the problem is patches to dracut will take eleven billion years to ship in RHCOS. Closes: #1950
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
#1946 (comment)
In #1946, we gave dracut
cap_mknodso it can create/dev/[u]randomso it's baked in the final CPIO (background on this is in https://bugzilla.redhat.com/show_bug.cgi?id=1778940). However, running the compose in a rootless podman will still fail tomknod. And because RHCOS now relies on those devices being there on early boot, we now fail to boot.We need to either bake it into the CPIO ourselves or (ideally) teach dracut to do this.
The text was updated successfully, but these errors were encountered: