-
Notifications
You must be signed in to change notification settings - Fork 2.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
cpu.cfs_quota_us: permission denied + mkdir issue on debian jessie #57
Comments
I think here are two different problems. I'm going to investigate the second problem with mkdir. Could you apply this patch and show output?
|
@avagin I'm getting the same error. Output after Patch: DEBU[0000] Apply configuration for net_prio
DEBU[0000] Apply configuration for perf_event
DEBU[0000] Apply configuration for devices
WARN[0000] signal: killed
FATA[0000] Container start failed: [8] System error: mkdir : no such file or directory edit: Missed the update to the container spec. Now the above Debug messages are gone. However, the following remains: WARN[0000] signal: killed
FATA[0000] Container start failed: [8] System error: mkdir : no such file or directory |
If i run it with this patch I get a few different outputs every other run:
At least those with the mkdir error seem to have in common that they all are applying the memory stuff before it crashes. |
Just a note: On my systems I didn't set the "cgroup_enable=memory swapaccount=1" options that seem to be required for memory limiting, maybe that has something to do with this problem. |
Could you attache output for
|
cgroup
mountinfo
|
You need to mount the memroy controller: |
As for cfs_quota_us, you probably use a kernel where this parameter isn't supported
|
Mounting using that command doesn't work.
Even if those two things aren't supported, runc shouldn't fail with an |
@lukas2511 GRUB_CMDLINE_LINUX="cgroup_enable=memory" Remember to update-grub2 afterwards |
@jkahrs |
@lukas2511 We have optional controllers (e.g. hugetlb), so I think it's possible to make the memory controller optional too, if memory limits are not set. |
I agree with @avagin . Memory shouldn't be mandatory. |
Fixes: opencontainers#57 Normally all cgroup subsystems are optional except device cgroup, but memory cgroup optional was broken by: docker-archive/libcontainer#637 This patch fixes this. Signed-off-by: Qiang Huang <h.huangqiang@huawei.com>
Normally all cgroup subsystems are optional except device cgroup, but memory cgroup optional was broken by: docker-archive/libcontainer#637 |
Closed in #118 |
I can't get this to run on my debian jessie machine(s) because of (seemingly) multiple issues.
I created rootfs with
docker export $(docker create busybox) > busybox.tar; mkdir rootfs; tar -C rootfs -xf busybox.tar
After creating container.json with
runc spec | sed 's/daemon/root/' > container.json
i get the following error runningrunc
:If i run it again the error changes:
runc --debug
output looks exactly the same. No more details.Running strace i can see that the code actually does try to use
mkdir("")
after reading an empty string from/proc/self/mountinfo
:If i remove the cpu (and memory) definitions from container.json i get that mkdir error directly.
The text was updated successfully, but these errors were encountered: