Detect support for fuse-overlayfs #198
Conversation
Signed-off-by: cgruver <[email protected]>
Signed-off-by: cgruver <[email protected]>
Signed-off-by: cgruver <[email protected]>
| @@ -5,6 +5,16 @@ if [ ! -d "${HOME}" ]; then | |||
| mkdir -p "${HOME}" | |||
| fi | |||
|
|
|||
| # Configure container builds to use vfs or fuse-overlayfs | |||
| if [ ! -d "${HOME}/.config/containers" ]; then | |||
There was a problem hiding this comment.
@dkwon17 with this change we should probably update the docs as well, right?
e.g. https://docs.redhat.com/en/documentation/red_hat_openshift_dev_spaces/3.17/html-single/user_guide/index?extIdCarryOver=true&sc_cid=701f2000001Css5AAC#enabling-overlay-with-a-configmap
Now UDI will be smart enough to detect fuse automatically
There was a problem hiding this comment.
@ibuziuk , yes, a configmap will not be needed anymore if the tooling container image is based on ubi/udi
|
verified on Developer Sandbox: |
| @@ -5,6 +5,16 @@ if [ ! -d "${HOME}" ]; then | |||
| mkdir -p "${HOME}" | |||
| fi | |||
|
|
|||
| # Configure container builds to use vfs or fuse-overlayfs | |||
| if [ ! -d "${HOME}/.config/containers" ]; then | |||
There was a problem hiding this comment.
What happens if persistent home is enabled and fuse-overlayfs gets disabled? Would ${HOME}/.config/containers/storage.conf need to be rewritten? Would the outer if block stop that from happening?
There was a problem hiding this comment.
The only way to disable fuse-overlayfs would be to apply a machine-config to the cluster, so I would say it's not likely. And if it did happen, that would be a disruptive change.
The reason that I put the first if block there is to not overwrite any existing podman config. The user might have added additional config to storage.conf which we would not want to clobber on a workspace restart.
There was a problem hiding this comment.
The if block could be a bit more targeted to look specifically for storage.conf I suppose.
if [ ! -f "${HOME}/.config/containers/storage.conf" ]; thenBut, there are valid configs where there is no storage.conf file, but other content under ${HOME}/.config/containers. Which is why I just check for the existence of the directory, not the file.
There was a problem hiding this comment.
Note: As of OCP 4.15 /dev/fuse is available for leaking into a container OOTB. No cluster change required.
With OCP 4.17 /dev/net/tun is also available.
Pod annotation: io.kubernetes.cri-o.Devices: "/dev/fuse,/dev/net/tun"
So, to disable fuse-overlayfs support would require disruptive intervention on the part of cluster admins.
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: cgruver, dkwon17, ibuziuk The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Add logic to
entrypoint.shto detect support forfuse-overlayfs