fix max open files limit for qemu

[ghost]

Does this PR introduce a user-facing change?

No.

- fix max open files limit for qemu (mainly applicable to MacOS, issue: https://github.com/containers/podman/issues/16106 ) 

Fixes issue: #16106

[rhatdan]

Thanks @protosam
/approve
LGTM
@baude @n1hility @mheon PTAL

[openshift-ci]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: protosam, rhatdan

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [rhatdan]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[packit-as-a-service]

Ephemeral COPR build failed. @containers/packit-build please check.

[n1hility]

Setting the limit this high can potentially fail. On Linux anything greater than /proc/sys/fs/nr_open can throw EPERM, and on Mac OS, some versions can through EINVAL if you exceed OPEN_MAX(10240). There could also be cases where the hard limit is set lower than the system max, in which you could raise the soft to match the hard limit. Most likely podman machine is not running as root, so that could be the solution to just set soft to match the current hard limit

[ghost]

Can't find solid documentation on it, so I'm guessing that syscall.Rlimit.Max is the hard limit and syscall.Rlimit.Cur is the soft limit.

Does the last change I made resolve this?

[n1hility]

Yes, looks good and thats right Cur = soft. FYI you can find the docs in the OSX/Linux man pages man setrlimit

[n1hility]

Thanks for the PR! Just one note above on the usage of RLIIMIT_INFINITY

[ghost]

Renamed the variable rLimit to rLimitNoFile to be less vague.

[rhatdan]

Compilation is failing.

[Luap99]

There is already code in cmd/podman/early_init_linux.go to handle this for linux for all podman processes.

I suggest you simply extend that file to also run on macos instead of setting this only for a single command.

[ghost]

There is already code in cmd/podman/early_init_linux.go to handle this for linux for all podman processes.

I suggest you simply extend that file to also run on macos instead of setting this only for a single command.

This looks like dead code to me. I don't see that setRLimits is being consumed in this repo.

% grep -ril setRLimits
./cmd/podman/early_init_linux.go
./vendor/github.com/containers/buildah/CHANGELOG.md
./vendor/github.com/containers/buildah/changelog.txt
./vendor/github.com/containers/buildah/chroot/run_common.go

[Luap99]

it is called in earlyInitHook() which is called from cobra.OnInitialize() in cmd/podman/root.go

[ghost]

Works for me. I've made changes to that affect. Let's see if this passes CI. 🤞

[ghost]

Not sure what to do about the "Build Each Commit" job. Should I just squash this commit history? Idk if it understand how to handle that based on it's output.

[n1hility]

Not sure what to do about the "Build Each Commit" job. Should I just squash this commit history? Idk if it understand how to handle that based on its output.

Yes please squash your commits. The rule is that if it's needed (make review easier) you can have multiple commits, but each individual commit has to be buildable and pass lint and whitespace checks on its own. Also, its good practice to keep the history of the project clean as possible: the commits should all build toward the same destination vs a local iterative work-stream where the changes replace and undo each other. What I recommend doing if you want to preserve your local history, is to just create a new branch as a local backup, and rebase the existing one to a clean patch set, and then force push. In this case it's a small enough change that one commit should be sufficient.

[TomSweeneyRedHat]

LGTM

[n1hility]

/lgtm

thanks again!