Summary:
rlimit support

Description:

1. Proposed title of this feature request

rlimit cgroups support for OpenShift Container Platform to enable to change the ulimit values

2. What is the nature and description of the request?

After creating Pods, the default size of “POSIX message queues” cannot be changed to a larger size with “ulimit -q” because Linux kernel doesn't have rlimit cgroup yet.
As an unsupported background, upstream Kubernetes has been discussing rlimit changes since 2015, but has not resolved them at this time.
With the subsequent definition of OCI runtime spec, Kubernetes commented that short-term solutions using OCI hook and the like could be applied even if they were not supported.
And most importantly it's part of the OCI runtime spec: https://github.com/opencontainers/runtime-spec/blob/master/config.md#posix-process'
And cri upstream mentions rlimit support.

3. Why does the customer need this? (List the business requirements here)

The customer wants OpenShift to support changes to increase posix message queues Pods. The reason is to implement their application.
・ If you do not change ulimit, pod/APP of the app can start Pod/APP, but call processing is not possible. This is because the message size for call processing according to the 5G RAN standard is larger than the queue size if you do not change ulimit.
・ In addition, we are concerned that multithreaded programming that deals with large message data other than just call control will be greatly limited, and the performance effect of the app will occur. As an example, logging is a separate thread from call control, with performance effects in take into account. When the amount of queuing here decreases, there is a large limitation on the log that can be retrieved. Alternatively, if you give up multithreading, you'll have a performance impact.