[Lxc-users] Problem with cgroup with LXc

Mon Nov 1 14:05:11 UTC 2010

Dne 26.10.2010 15:03, Daniel Lezcano napsal(a):
> On 10/26/2010 08:38 AM, Miroslav Lednicky, AVONET, s.r.o. wrote:
>> Hello all,
>>
>> I started using LXCs. They are very nice.
>> But I have problems with cgroup.
>>
>> There are problem with remove old informations from cgroup subdirectory.
>> It happend only sometimes. Typicaly zabbix agent in LXC can generate
>> this problem. But not always.
>>
>> Please see:
>>
>> ls -l /cgroup/test_lxc
>>
>> drwxr-xr-x 3 root root 0 2010-09-29 23:07 10194
>> drwxr-xr-x 3 root root 0 2010-10-01 21:11 11382
>> drwxr-xr-x 3 root root 0 2010-10-03 18:29 12632
>> drwxr-xr-x 3 root root 0 2010-09-15 15:10 1715
>> drwxr-xr-x 3 root root 0 2010-10-15 07:31 20270
>> drwxr-xr-x 3 root root 0 2010-10-16 02:05 20468
>> drwxr-xr-x 3 root root 0 2010-10-16 22:42 21090
>> drwxr-xr-x 3 root root 0 2010-10-19 04:58 22349
>> drwxr-xr-x 3 root root 0 2010-08-27 16:09 22455
>> drwxr-xr-x 3 root root 0 2010-08-29 10:45 23636
>> drwxr-xr-x 3 root root 0 2010-09-16 19:10 2398
>> drwxr-xr-x 3 root root 0 2010-10-22 00:27 24182
>> drwxr-xr-x 3 root root 0 2010-10-26 06:45 27044
>> drwxr-xr-x 3 root root 0 2010-09-04 18:26 27119
>> drwxr-xr-x 3 root root 0 2010-09-05 04:24 27187
>> drwxr-xr-x 3 root root 0 2010-09-09 21:39 30581
>> drwxr-xr-x 3 root root 0 2010-09-20 10:10 4793
>> -r--r--r-- 1 root root 0 2010-08-02 13:53 cgroup.procs
>> -r--r--r-- 1 root root 0 2010-08-02 13:53 cpuacct.stat
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuacct.usage
>> -r--r--r-- 1 root root 0 2010-08-02 13:53 cpuacct.usage_percpu
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpu.rt_period_us
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpu.rt_runtime_us
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.cpu_exclusive
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.cpus
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.mem_exclusive
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.mem_hardwall
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.memory_migrate
>> -r--r--r-- 1 root root 0 2010-08-02 13:53 cpuset.memory_pressure
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.memory_spread_page
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.memory_spread_slab
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.mems
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.sched_load_balance
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpuset.sched_relax_domain_level
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 cpu.shares
>> --w------- 1 root root 0 2010-08-02 13:53 devices.allow
>> --w------- 1 root root 0 2010-08-02 13:53 devices.deny
>> -r--r--r-- 1 root root 0 2010-08-02 13:53 devices.list
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 freezer.state
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 memory.failcnt
>> --w------- 1 root root 0 2010-08-02 13:53 memory.force_empty
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 memory.limit_in_bytes
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 memory.max_usage_in_bytes
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 memory.memsw.failcnt
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 memory.memsw.limit_in_bytes
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 memory.memsw.max_usage_in_bytes
>> -r--r--r-- 1 root root 0 2010-08-02 13:53 memory.memsw.usage_in_bytes
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 memory.soft_limit_in_bytes
>> -r--r--r-- 1 root root 0 2010-08-02 13:53 memory.stat
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 memory.swappiness
>> -r--r--r-- 1 root root 0 2010-08-02 13:53 memory.usage_in_bytes
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 memory.use_hierarchy
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 net_cls.classid
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 notify_on_release
>> -rw-r--r-- 1 root root 0 2010-08-02 13:53 tasks
>>
>> ls -R1 10194
>> 10194:
>> 2
>> cgroup.procs
>> cpuacct.stat
>> cpuacct.usage
>> cpuacct.usage_percpu
>> cpu.rt_period_us
>> cpu.rt_runtime_us
>> cpuset.cpu_exclusive
>> cpuset.cpus
>> cpuset.mem_exclusive
>> cpuset.mem_hardwall
>> cpuset.memory_migrate
>> cpuset.memory_pressure
>> cpuset.memory_spread_page
>> cpuset.memory_spread_slab
>> cpuset.mems
>> cpuset.sched_load_balance
>> cpuset.sched_relax_domain_level
>> cpu.shares
>> devices.allow
>> devices.deny
>> devices.list
>> freezer.state
>> memory.failcnt
>> memory.force_empty
>> memory.limit_in_bytes
>> memory.max_usage_in_bytes
>> memory.memsw.failcnt
>> memory.memsw.limit_in_bytes
>> memory.memsw.max_usage_in_bytes
>> memory.memsw.usage_in_bytes
>> memory.soft_limit_in_bytes
>> memory.stat
>> memory.swappiness
>> memory.usage_in_bytes
>> memory.use_hierarchy
>> net_cls.classid
>> notify_on_release
>> tasks
>>
>> 10194/2:
>> cgroup.procs
>> cpuacct.stat
>> cpuacct.usage
>> cpuacct.usage_percpu
>> cpu.rt_period_us
>> cpu.rt_runtime_us
>> cpuset.cpu_exclusive
>> cpuset.cpus
>> cpuset.mem_exclusive
>> cpuset.mem_hardwall
>> cpuset.memory_migrate
>> cpuset.memory_pressure
>> cpuset.memory_spread_page
>> cpuset.memory_spread_slab
>> cpuset.mems
>> cpuset.sched_load_balance
>> cpuset.sched_relax_domain_level
>> cpu.shares
>> devices.allow
>> devices.deny
>> devices.list
>> freezer.state
>> memory.failcnt
>> memory.force_empty
>> memory.limit_in_bytes
>> memory.max_usage_in_bytes
>> memory.memsw.failcnt
>> memory.memsw.limit_in_bytes
>> memory.memsw.max_usage_in_bytes
>> memory.memsw.usage_in_bytes
>> memory.soft_limit_in_bytes
>> memory.stat
>> memory.swappiness
>> memory.usage_in_bytes
>> memory.use_hierarchy
>> net_cls.classid
>> notify_on_release
>> tasks
>>
>> It looks like as this problem:
>>
>> http://www.mail-archive.com/devel@openvz.org/msg19736.html
>>
>> But I have no solution.
>>
>> Can somebody help me?
>>
>> It is big problem with lxc-stop and lx-start. I must restarting server.
>
> Weird.
>
> Is it possible the zabbix application creates a new namespace ?

T thing that no. But in logs is sometime:

warning: process `zabbix_agentd' used the deprecated sysctl system call 
with 1.55.

I running zabbix agent in most of my containers, but i have problem
only with one.

I stoped zabbix agent, but there are some numeric directories in
container cgroup.

But not too much. 11 directories after one month running.

I cannot stop and start LXC without restart server now.

> If it is
> the case, it's a known problem with the ns_cgroup, that has been solved
> for the next kernel release because the ns_cgroup won't be used anymore,
> so you the application can unshare whatever they want.
> Also, is it possible, you didn't pay attention and you tried to launch a
> container inside a container ?

No.

Thank you and best regards,

Miroslav.

-- 
Miroslav Lednicky, AVONET, s.r.o.