Issues with Debian Wheezy about lmctfy HOT 14 CLOSED

is memcg compiled into the kernel? Can you manually mount a cgroupfs with
-o memory ?

On Wed, Oct 9, 2013 at 11:35 AM, Nat Welch [email protected] wrote:

$ sudo lmctfy init "

cgroup_mount:{
mount_path:'/dev/cgroup/cpu'
hierarchy:CGROUP_CPU hierarchy:CGROUP_CPUACCT
}
cgroup_mount:{
mount_path:'/dev/cgroup/cpuset' hierarchy:CGROUP_CPUSET
}
cgroup_mount:{
mount_path:'/dev/cgroup/freezer' hierarchy:CGROUP_FREEZER
}
cgroup_mount:{
mount_path:'/dev/cgroup/memory' hierarchy:CGROUP_MEMORY
}"
Command exited with error message: 9: Failed to mount hierarchy with ID "memory" at "/dev/cgroup/memory"

$ lmctfy -v
lmctfy version 0.1-0

$ uname -a
Linux byok 3.2.0-4-amd64 #1 SMP Debian 3.2.46-1 x86_64 GNU/Linux

I'm running on Google Compute Engine. ([email protected] if you wanna
talk off bug about how to replicate this).

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/8
.

from lmctfy.

The output of /proc/cgroups may be useful as well.

from lmctfy.

It's default debian kernel.

 lsmod
Module                  Size  Used by
fuse                   62020  0 
btrfs                 505636  0 
libcrc32c              12426  1 btrfs
zlib_deflate           25638  1 btrfs
ufs                    58774  0 
qnx4                   13184  0 
hfsplus                71616  0 
hfs                    45877  0 
minix                  27623  0 
ntfs                  163882  0 
vfat                   17316  0 
msdos                  17077  0 
fat                    45642  2 msdos,vfat
jfs                   137196  0 
xfs                   590895  0 
reiserfs              192132  0 
ext3                  162072  0 
jbd                    56902  1 ext3
ext2                   59231  0 
efivars                17916  0 
dm_mod                 63645  0 
snd_pcsp               13742  0 
snd_pcm                68083  1 snd_pcsp
snd_page_alloc         13003  1 snd_pcm
snd_timer              22917  1 snd_pcm
crc32c_intel           12747  1 
i2c_piix4              12536  0 
snd                    52889  3 snd_timer,snd_pcm,snd_pcsp
ghash_clmulni_intel    13173  0 
evdev                  17562  1 
processor              28157  0 
soundcore              13065  1 snd
thermal_sys            18040  1 processor
i2c_core               23876  1 i2c_piix4
aesni_intel            50667  0 
aes_x86_64             16843  1 aesni_intel
aes_generic            33026  2 aes_x86_64,aesni_intel
button                 12937  0 
cryptd                 14517  2 aesni_intel,ghash_clmulni_intel
ext4                  350763  1 
crc16                  12343  1 ext4
jbd2                   62115  1 ext4
mbcache                13114  3 ext4,ext2,ext3
sg                     25874  0 
sd_mod                 36136  2 
crc_t10dif             12348  1 sd_mod
virtio_scsi            13319  1 
virtio_net             17808  0 
scsi_mod              162269  3 virtio_scsi,sd_mod,sg
virtio_pci             13207  0 
virtio_ring            13057  3 virtio_pci,virtio_net,virtio_scsi
virtio                 13093  4 virtio_ring,virtio_pci,virtio_net,virtio_scsi

$ cat  /proc/cgroups
#subsys_name    hierarchy   num_cgroups enabled
cpuset  2   1   1
cpu 1   1   1
cpuacct 1   1   1
memory  0   1   0
devices 0   1   1
freezer 3   1   1
net_cls 0   1   1
blkio   0   1   1
perf_event  0   1   1

I don't know enough about cgroups. How do I manually mount one?

from lmctfy.

To mount, as root: mount -t cgroup none {some dir to mount to} -o memory

from lmctfy.

http://cl.natw.me/Rr0L is the build config file that was in /boot.

from lmctfy.

root@byok:~# mount -t cgroup none /mnt -o memory
mount: special device none does not exist

from lmctfy.

What is currently mounted btw? Can you run: cat /proc/mounts | grep cgroup

from lmctfy.

You should give him a full command :)

thockin@flaptop:~$ mkdir /tmp/mem

thockin@flaptop:~$ sudo mount -t cgroup cgroup /tmp/mem -o memory
[sudo] password for thockin:

thockin@flaptop:~$ ls /tmp/mem
cgroup.clone_children memory.move_charge_at_immigrate
cgroup.event_control memory.numa_stat
cgroup.procs memory.oom_control
memory.failcnt memory.soft_limit_in_bytes
memory.force_empty memory.stat
memory.limit_in_bytes memory.swappiness
memory.max_usage_in_bytes memory.usage_in_bytes
memory.memsw.failcnt memory.use_hierarchy
memory.memsw.limit_in_bytes notify_on_release
memory.memsw.max_usage_in_bytes release_agent
memory.memsw.usage_in_bytes tasks

On Wed, Oct 9, 2013 at 11:48 AM, Victor Marmol [email protected]:

To mount, as root: mount -t cgroup none -o memory

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/8#issuecomment-25997582
.

from lmctfy.

$ mkdir /tmp/mem; sudo mount -t cgroup cgroup /tmp/mem -o memory;  ls /tmp/mem
mount: special device cgroup does not exist

That sounds like I don't have cgroup support?

But this looks slightly more promising:

$ cat /proc/mounts | grep cgroup
cgroup /dev/cgroup/cpu cgroup rw,relatime,cpuacct,cpu 0 0
cgroup /dev/cgroup/cpuset cgroup rw,relatime,cpuset 0 0
cgroup /dev/cgroup/freezer cgroup rw,relatime,freezer 0 0

from lmctfy.

Seems like it doesn't have support for memcg. From the config it looks like memcg is disabled:

CONFIG_CGROUP_MEM_RES_CTLR=y
CONFIG_CGROUP_MEM_RES_CTLR_DISABLED=y

and it looks like Docker reached a similar conclusions with the default Debian kernel:

moby/moby#396

from lmctfy.

Try a reboot with "cgroup_enable=memory" on your kernel commandline?

I had to go google the meaning of the _DISABLED param, but that seems tobe
what it means. Compiled in, but disabled by default.

On Wed, Oct 9, 2013 at 11:59 AM, Victor Marmol [email protected]:

Seems like it doesn't have support for memcg. From the config it looks
like memcg is disabled:

CONFIG_CGROUP_MEM_RES_CTLR=y
CONFIG_CGROUP_MEM_RES_CTLR_DISABLED=y

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/8#issuecomment-25998428
.

from lmctfy.

Awesome, I rebooted with that option (as described here: https://wiki.debian.org/LXC#Prepare_the_host) and it worked!

natwelch@byok:~$ sudo lmctfy init "
>   cgroup_mount:{
>     mount_path:'/dev/cgroup/cpu'
>     hierarchy:CGROUP_CPU hierarchy:CGROUP_CPUACCT
>   }
>   cgroup_mount:{
>     mount_path:'/dev/cgroup/cpuset' hierarchy:CGROUP_CPUSET
>   }
>   cgroup_mount:{
>     mount_path:'/dev/cgroup/freezer' hierarchy:CGROUP_FREEZER
>   }
>   cgroup_mount:{
>     mount_path:'/dev/cgroup/memory' hierarchy:CGROUP_MEMORY
>   }"
natwelch@byok:~$  sudo lmctfy create test "memory:{limit:100000000}"
natwelch@byok:~$ sudo lmctfy list containers
name="/test"
natwelch@byok:~$ sudo lmctfy run test "echo hello world"
hello world
exit_code="0"

Thanks guys!

from lmctfy.

Could you please add this knowledge to the documentation. As it is know spread in a closed issue is hard to find.

from lmctfy.

mars@node1:/community/cookbooks$ sudo mount -t cgroup -o debug cgroup /cgroup
[sudo] password for mars:
mount: mount point /cgroup does not exist
mars@node1:/community/cookbooks$ sudo mkdir /cgroup
mars@node1:~/community/cookbooks$ sudo mount -t cgroup -o debug cgroup /cgroup
mount: special device cgroup does not exist

looks like it has no cgroup support, how to enable cgroup for operating system ?

from lmctfy.