关于remount_lxcfs的这个功能

[markzhang0928]

最近的这个remount_lxcfs 是不是还没有实现哦，我看codecov上的代码还是空的。https://codecov.io/gh/alibaba/pouch/pull/885/diff?src=pr&el=tree#diff-Y2xpL21haW4uZ28=

[CodeJuan]

@markzhang0928 是的，这一块还没有实现。方法就是

1. 将宿主机上lxcfs的上一层目录挂到容器里

-v /var/lib/lxc/:/var/lib/lxc/:shared  \
-v /var/lib/lxc/lxcfs/proc/uptime:/proc/uptime \
-v /var/lib/lxc/lxcfs/proc/swaps:/proc/swaps  \
-v /var/lib/lxc/lxcfs/proc/stat:/proc/stat  \
-v /var/lib/lxc/lxcfs/proc/diskstats:/proc/diskstats \
-v /var/lib/lxc/lxcfs/proc/meminfo:/proc/meminfo \
-v /var/lib/lxc/lxcfs/proc/cpuinfo:/proc/cpuinfo

2. 在systemd的增加ExecStartPost=remount_lxcfs
3. 在lxcfs进程重启之后，在容器里再mount一次mount -B "/var/lib/lxc/lxcfs/proc/$file" "/proc/$file"

[markzhang0928]

@CodeJuan thx～

但是我试过你在这里提到的方法，用nsenter 进行mount -B.
报错是：mount: mount point /proc/meminfo does not exist.

宿主上lxcfs进程重启后，容器中的/proc/$file文件会是这样

root@5d9039f89f8e:/# ls -la /proc/
ls: cannot access /proc/diskstats: Transport endpoint is not connected
ls: cannot access /proc/swaps: Transport endpoint is not connected
ls: cannot access /proc/uptime: Transport endpoint is not connected
ls: cannot access /proc/stat: Transport endpoint is not connected
ls: cannot access /proc/meminfo: Transport endpoint is not connected
ls: cannot access /proc/cpuinfo: Transport endpoint is not connected
...
-?????????   ? ?    ?       ?            ? cpuinfo
-?????????   ? ?    ?       ?            ? meminfo
-?????????   ? ?    ?       ?            ? stat
-?????????   ? ?    ?       ?            ? swaps
-?????????   ? ?    ?       ?            ? uptime

OS:

Distributor ID:	Debian
Description:	Debian GNU/Linux 8.6 (jessie)
Release:	8.6
Codename:	jessie

Kernel:

Linux 3.16.0-4-amd64 #1 SMP Debian 3.16.36-1+deb8u1 (2016-09-03) x86_64 GNU/Linux

refs:
AliyunContainerService/pouch#140
lxc/lxcfs#193

[markzhang0928]

奥, 最近回顾了一下，找到原因了。 没有mount /var/lib/lxc 到容器的mount ns.

[JaeGerW2016]

lxcfs能不能以daemonset的形式在节点运行，ExecStartPost=remount_lxcfs用postStart的pod hook实现remount？

[markzhang0928]

lxcfs能不能以daemonset的形式在节点运行，ExecStartPost=remount_lxcfs用postStart的pod hook实现remount？

remount 是可以，前提是容器里面有mount point.

[julyclyde]

求教，我用nsenter mount -B就可以成功；

docker exec进去mount -B就提示说
mount point is not a directory

这是为什么呢？

[ganker-z]

奥, 最近回顾了一下，找到原因了。 没有mount /var/lib/lxc 到容器的mount ns.

请问下，怎么mount /var/lib/lxc 到容器的mount ns

[zxq1002]

奥, 最近回顾了一下，找到原因了。 没有mount /var/lib/lxc 到容器的mount ns.

请问下，怎么mount /var/lib/lxc 到容器的mount ns

同问，把宿主机的/var/lib/lxc mount到容器的/var/lib/lxc？

[JaeGerW2016]

邮件已收到，我会尽快查看并回复

[fjibj]

还是不行，请帮忙看看 @CodeJuan @markzhang0928 @JaeGerW2016

lxcfs是在宿主机上手工启停的
# /usr/bin/lxcfs -l /var/lib/lxc/lxcfs/

容器启动命令：
`docker run -dt -m 5g \

-v=/sys/fs/cgroup:/sys/fs/cgroup:ro \

-v=/var/lib/lxc/:/var/lib/lxc/:shared \

-v=/var/lib/lxc/lxcfs/proc/diskstats:/proc/diskstats:rw,rslave \

-v=/var/lib/lxc/lxcfs/proc/meminfo:/proc/meminfo:rw,rslave \

-v=/var/lib/lxc/lxcfs/proc/stat:/proc/stat:rw,rslave \

-v=/var/lib/lxc/lxcfs/proc/swaps:/proc/swaps:rw,rslave \

-v=/var/lib/lxc/lxcfs/proc/uptime:/proc/uptime:rw,rslave \

-v=/var/lib/lxc/lxcfs/proc/loadavg:/proc/loadavg:rw,rslave \

--name ddd \

XXXXX`

在宿主机上执行：
`# nsenter --target 14591 --mount -- mount -B /var/lib/lxcfs/proc/meminfo /proc/meminfo

mount: mount point /proc/meminfo does not exist`

在容器内执行：
`[root@master lxcfs]# mount -B "/var/lib/lxc/lxcfs/proc/cpuinfo" "/proc/cpuinfo"

mount: mount point /proc/cpuinfo is not a directory`

[JaeGerW2016]

前提是容器里面有mount point @fjibj

[fjibj]

怎么设置容器里面的mount point？

我在docker run时已经把/var/lib/lxc/挂进去了，
-v=/var/lib/lxc/:/var/lib/lxc/:shared \
还需要做其他设置吗？
@JaeGerW2016 @markzhang0928