Container Memory Usage Always 0 in 4.3.0

[robd003]

• I have tried with the latest version of Docker Desktop
• I have tried disabling enabled experimental features
• I have uploaded Diagnostics
• Diagnostics ID: 17C272C8-850F-4CEB-AE3C-9BC16229A0B0/20211203180843 -- Virtualization.framework
• Diagnostics ID: 17C272C8-850F-4CEB-AE3C-9BC16229A0B0/20211204031030 -- Hypervisor.framework

Expected behavior

Show memory usage of containers

Actual behavior

Containers always show 0 bytes used

Information

• macOS Version: 12.1
• Intel chip or Apple chip: M1 Max
• Docker Desktop Version: 4.3.0

Steps to reproduce the behavior

Start any container. All containers report 0 for memory usage. CPU, network and disk I/O are all reported correctly.

[robd003]

I've tried using Virtualization.framework and Virtualization.framework and both don't report memory usage.

Oddly if I use the cli docker stats container_name I can see the memory stats, but in the dashboard it's always 0

[robd003]

Confirmed that this is broken on Intel Macs too.

It worked fine in 4.2.0, but broke with the 4.3.0 release.

[Avertae]

More input on this, in case containerized app depends on "/sys/fs/cgroup/memory/memory.limit_in_bytes" it fails to start with:
cat: /sys/fs/cgroup/memory/memory.limit_in_bytes: No such file or directory

4.2.0 was ok.

[stephen-turner]

Thanks for the report, we're looking into this.

[stephen-turner]

Also broken on Windows Hyper-V but not Windows WSL 2. We've added a release note and we're working on a fix.

[mat007]

We have now fixed this issue in Docker Desktop 4.4.2, thanks!

[robd003]

@mat007 Awesome! Upgrading right now to verify

[browdues]

Hey there, I noticed this post today and upgraded to docker desktop 4.4.2, per the recommendation.

However, I still don't see both the mem and the cpu directories.

Note that I am on Mac, Catalina OS, using an Intel I9.

Here is what I see when I examine the contents of /sys/fs/cgroup:

root@eaas-configrun-service-7cbc8744ff-7zvgf:/# cd /sys/fs/cgroup/
cgroup.controllers        cgroup.threads            cpuset.cpus.partition     hugetlb.1GB.rsvd.max      io.latency                memory.max                pids.current
cgroup.events             cgroup.type               cpuset.mems               hugetlb.2MB.current       io.max                    memory.min                pids.events
cgroup.freeze             cpu.max                   cpuset.mems.effective     hugetlb.2MB.events        io.stat                   memory.oom.group          pids.max
cgroup.max.depth          cpu.stat                  hugetlb.1GB.current       hugetlb.2MB.events.local  memory.current            memory.stat               rdma.current
cgroup.max.descendants    cpu.weight                hugetlb.1GB.events        hugetlb.2MB.max           memory.events             memory.swap.current       rdma.max
cgroup.procs              cpu.weight.nice           hugetlb.1GB.events.local  hugetlb.2MB.rsvd.current  memory.events.local       memory.swap.events        
cgroup.stat               cpuset.cpus               hugetlb.1GB.max           hugetlb.2MB.rsvd.max      memory.high               memory.swap.high          
cgroup.subtree_control    cpuset.cpus.effective     hugetlb.1GB.rsvd.current  io.bfq.weight             memory.low                memory.swap.max     

Is this expected or is this still broken? In 4.2 (and prior versions) we still saw both the cpu and mem directories.

[Avertae]

Fix is not enabled OOB, which was blamed already here
TLDR;
~/Library/Group Containers/group.com.docker/settings.json
"deprecatedCgroupv1": true,

[browdues]

Thank you

[docker-robott]

Closed issues are locked after 30 days of inactivity.
This helps our team focus on active issues.

If you have found a problem that seems similar to this, please open a new issue.

Send feedback to Docker Community Slack channels #docker-for-mac or #docker-for-windows.
/lifecycle locked