gosigar can read total memory in many system. (include cgroup)
can we use it for support other system?

Proposal

Add system memory as an available provider in order to avoid OOMs when running outside of cgroups.

For workloads running outside of cgroups, or for systems without cgroups, it would be useful to have a provider that exposes the total system memory. This way you cold set a percent of system memory automatically as the limit.

If it works normally, does automemlimit work without being affected by the resource with just the code of the example regardless of the size of the resource? I'm currently using it by setting it in the environment variable, but I'm not sure yet because it's been a while since the latest version.

Hey, I just started using this package. I like it! But, I use it alongside automaxprocs, and I noticed automaxprocs always logs when it changes the GOMAXPROCS setting, whereas your package only logs when debug mode is enabled. I think since most people will use both packages, it would be nice if you also logged by default.

Proposal

In order to support different logging libraries, it would be nice to have logging be an interface to something compatible with the current stdlib log package.

For example:

import "log"

type Logger interface {
	Fatalf(format string, v ...interface{})
	Errorf(format string, v ...interface{})
	Warnf(format string, v ...interface{})
	Infof(format string, v ...interface{})
	Debugf(format string, v ...interface{})
}

Can you clarify whether it is required to put this code in an init() method?

I have deployed a k8s pod with both automemlimit and automaxproc and I have set the resources to the following:

resources:
  limits:
    cpu: 2
    memory: 4Gi

  requests:
     cpu: 1
     memory: 2Gi

automemlimits uses the request value rather than the limits values set on the k8s manifest. Is this expected behavior?

automaxprocs uses the limits value set on manifest rather than the request value.

I was under the impression that automemlimits did the same thing for memory limits like automaxprocs did for cpu limits.

Am I missing something here?

Hey, I'd also be interested in this. Well, not necessarily sigar, but rather having automemlimit "just work" even outside a container.

For my use case, I am using Google Cloud run, and currently automemlimit doesn't work because it can't see it is running in a container. Specifically, I get the message

2023/04/23 01:40:07 failed to set GOMEMLIMIT: open /sys/fs/cgroup/cgroup.controllers: no such file or directory.

Originally posted by @mbyio in #4 (comment)

In Kubernetes 1.27, there is now in-place pod resizing.

It would be useful to have a refresh interval that can update the memlimit in case the cgroup settings change.