New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
rpmostreepayload: Rework binds to be recursive #903
Conversation
Our hardcoded list of mount points missed critical ones like `/sys/firmware/efi/efivars`, which broke installation on EFI. I started to add that to the list, but then I realized there's a much simpler solution - make the binds recursive and let `mount` do the work for us. This is a bit more risky of a change, but it'll clearly be more maintainable long term. Down the line...we may want to simply walk over all the toplevel mount points in `/mnt/sysimage`, but I doubt anyone is going to add anything new that's not a subdirectory of `/sys`. See https://pagure.io/atomic-wg/issue/185 Related: https://github.com/rhinstaller/anaconda/issues/900
2d1fb8b
to
543e100
Compare
Here's an
|
Jenkins, it's ok to test. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good to me.
Looks good to me. |
can we get this merged? - we'd like to get things fixed so our next two week release of Atomic Host has a fix for this. |
can we get this merged in for f25 and cut a new rpm for Fedora? |
In rhinstaller#903 we changed all of rpmostreepayload's internal mounts to be recursive to fix `/boot/efi`. And indeed all should be recursive, except one - the `/sysroot` mount. By asking for that to be recursive we've created a loop. I noticed a while ago that the output of `findmnt` during an install was insane (the deployment directory was visible twice, etc.), but never debugged it. It turns out this was the cause. Fixing this makes the mounts look a lot more sane, and will help for future work I have for `/var` mounts.
In rhinstaller#903 we changed all of rpmostreepayload's internal mounts to be recursive to fix `/boot/efi`. And indeed all should be recursive, except one - the `/sysroot` mount. By asking for that to be recursive we've created a loop. I noticed a while ago that the output of `findmnt` during an install was insane (the deployment directory was visible twice, etc.), but never debugged it. It turns out this was the cause. Fixing this makes the mounts look a lot more sane, and will help for future work I have for `/var` mounts.
In rhinstaller#903 we changed all of rpmostreepayload's internal mounts to be recursive to fix `/boot/efi`. And indeed all should be recursive, except one - the `/sysroot` mount. By asking for that to be recursive we've created a loop. I noticed a while ago that the output of `findmnt` during an install was insane (the deployment directory was visible twice, etc.), but never debugged it. It turns out this was the cause. Fixing this makes the mounts look a lot more sane, and will help for future work I have for `/var` mounts.
Our hardcoded list of mount points missed critical ones
like
/sys/firmware/efi/efivars
, which broke installationon EFI.
I started to add that to the list, but then I realized there's a much
simpler solution - make the binds recursive and let
mount
do thework for us. This is a bit more risky of a change, but it'll
clearly be more maintainable long term.
Down the line...we may want to simply walk over all the toplevel mount
points in
/mnt/sysimage
, but I doubt anyone is going to add anythingnew that's not a subdirectory of
/sys
.See https://pagure.io/atomic-wg/issue/185