glusterfs containers can't start on 3.7 with docker 1.13+

[dustymabe]

In Fedora we currently have docker-1.13.1-42.git4402c09.fc27.x86_64. There was a change in kube 1.7 to use shared PID namespaces in a pod if docker > 1.13+. It causes issues with our gluster pods (systemd in a container):

Couldn't find an alternative telinit implementation to spawn.

This specific issue with gluster is documented upstream and they actually reverted the change (so back to default to no shared PID namespace) in kube 1.8. Kube 1.7 still has the behavior but there is a --docker-disable-shared-pid flag that we can give to the kubelet to change the behavior to what we desire (and what is the default in 1.6 & 1.8) and get glusterfs working again.

so... Two options for a fix:

• adding the --docker-disable-shared-pid flag to the kubelet using openshift ansible when origin 3.7 && docker 1.13
• backporting the patch from kube 1.8 into origin 1.7 that reverts the change.

Note: If we go the --docker-disable-shared-pid they'll probably remove that flag in the future when it is no longer needed so we might have to add logic to "remove" the flag on an upgrade one day. Upstream issue to track that here.

[sdodson]

@dustymabe I assume you have no reasonable way to downgrade to docker-1.12 on this as it's atomic host?
@sjenning assuming ^^ is "no" what should we tell folks who are in the situation where docker-1.13 is the only version of docker available?

[dustymabe]

@dustymabe I assume you have no reasonable way to downgrade to docker-1.12 on this as it's atomic host?

We've had docker 1.13 in Fedora for some time now and going backwards is not desirable.

@sjenning assuming ^^ is "no" what should we tell folks who are in the situation where docker-1.13 is the only version of docker available?

hi @sjenning :)

[sdodson]

Even if it weren't atomic host Docker 1.12 was never available for Fedora 27.

[sjenning]

PR opened to backport the revert from upstream openshift/origin#17722

[openshift-bot]

Issues go stale after 90d of inactivity.

Mark the issue as fresh by commenting /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.
Exclude this issue from closing by commenting /lifecycle frozen.

If this issue is safe to close now please do so with /close.

/lifecycle stale

[openshift-bot]

Stale issues rot after 30d of inactivity.

Mark the issue as fresh by commenting /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.
Exclude this issue from closing by commenting /lifecycle frozen.

If this issue is safe to close now please do so with /close.

/lifecycle rotten
/remove-lifecycle stale

[openshift-bot]

Rotten issues close after 30d of inactivity.

Reopen the issue by commenting /reopen.
Mark the issue as fresh by commenting /remove-lifecycle rotten.
Exclude this issue from closing again by commenting /lifecycle frozen.

/close

[openshift-ci-robot]

@openshift-bot: Closing this issue.

In response to this:

Rotten issues close after 30d of inactivity.

Reopen the issue by commenting /reopen.
Mark the issue as fresh by commenting /remove-lifecycle rotten.
Exclude this issue from closing again by commenting /lifecycle frozen.

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.