Yes, commenting out 'return -1' makes it starts and it seems to behave correctly.
The /proc/self/mountinfo done from inside the container is empty (as no proc is mounted). If I mount it (by lxc.mount.auto or directly inside the container) then: 42 20 0:2 /containers/mycnt/rootfs / rw - rootfs rootfs rw,size=90972k,nr_inodes=22743 43 42 0:19 / /dev rw,nodev,relatime - tmpfs none rw,size=100k,mode=755,uid=100000,gid=100000 44 43 0:5 /null /dev/null rw,relatime - devtmpfs devtmpfs rw,size=90972k,nr_inodes=22743,mode=755 45 43 0:5 /zero /dev/zero rw,relatime - devtmpfs devtmpfs rw,size=90972k,nr_inodes=22743,mode=755 46 43 0:5 /full /dev/full rw,relatime - devtmpfs devtmpfs rw,size=90972k,nr_inodes=22743,mode=755 47 43 0:5 /urandom /dev/urandom rw,relatime - devtmpfs devtmpfs rw,size=90972k,nr_inodes=22743,mode=755 48 43 0:5 /random /dev/random rw,relatime - devtmpfs devtmpfs rw,size=90972k,nr_inodes=22743,mode=755 49 43 0:5 /tty /dev/tty rw,relatime - devtmpfs devtmpfs rw,size=90972k,nr_inodes=22743,mode=755 50 43 0:5 /console /dev/console rw,relatime - devtmpfs devtmpfs rw,size=90972k,nr_inodes=22743,mode=755 51 50 0:11 /0 /dev/console rw,relatime - devpts devpts rw,gid=5,mode=620,ptmxmode=000 52 42 0:2 /shared /tmp rw - rootfs rootfs rw,size=90972k,nr_inodes=22743 30 43 0:20 / /dev/pts rw,relatime - devpts devpts rw,gid=100005,mode=620,ptmxmode=666 31 43 0:20 /0 /dev/tty1 rw,relatime - devpts devpts rw,gid=100005,mode=620,ptmxmode=666 32 42 0:18 / /proc rw,nodev,relatime - proc proc rw BTW. Just to be exact, in first mount log before starting lxc I forgot to enable cloning, so the exact line is: 19 18 0:16 / /sys/fs/cgroup rw,relatime - cgroup cgroup rw,cpuset,cpuacct,memory,devices,freezer,net_cls,clone_children instead of: 19 18 0:16 / /sys/fs/cgroup rw,relatime - cgroup cgroup rw,cpuset,cpuacct,memory,devices,freezer,net_cls On 29.07.2015 15:32, Serge Hallyn wrote: > If you rebuild lxc patching that paragraph to warn but not return > failure when the umount of './proc' fails, does the container start > properly? What does /proc/self/mountinfo inside the container then > look like? > > -serge > > Quoting Przemyslaw Rudy ([email protected]): >> /proc/self/mountinfo before the lxc-start: >> >> 1 1 0:2 / / rw - rootfs rootfs rw,size=90972k,nr_inodes=22743 >> 13 1 0:5 / /dev rw,relatime - devtmpfs devtmpfs >> rw,size=90972k,nr_inodes=22743,mode=755 >> 14 1 0:4 / /proc rw,relatime - proc proc rw >> 15 13 0:11 / /dev/pts rw,relatime - devpts devpts >> rw,gid=5,mode=620,ptmxmode=000 >> 16 13 0:13 / /dev/shm rw,relatime - tmpfs tmpfs rw,mode=777 >> 17 1 0:14 / /tmp rw,relatime - tmpfs tmpfs rw >> 18 1 0:15 / /sys rw,relatime - sysfs sysfs rw >> 19 18 0:16 / /sys/fs/cgroup rw,relatime - cgroup cgroup >> rw,cpuset,cpuacct,memory,devices,freezer,net_cls >> >> >> On 07/29/2015 04:07 AM, Serge Hallyn wrote: >>> I suspect you're right and we should simply ignore the failure. >>> >>> Can you show /proc/self/mountinfo before the lxc-start? >>> >>> Quoting Przemyslaw Rudy ([email protected]): >>>> -1 is rv, errno is 22 EINVAL >>>> >>>> The log: >>>> lxc-start 40.086 INFO lxc_start_ui - >>>> lxc_start.c:main:264 - using rcfile /containers/mycnt/config >>>> lxc-start 40.087 INFO lxc_confile - >>>> confile.c:config_idmap:1376 - read uid map: type u nsid 0 hostid 100000 >>>> range 65536 >>>> lxc-start 40.087 INFO lxc_confile - >>>> confile.c:config_idmap:1376 - read uid map: type g nsid 0 hostid 100000 >>>> range 65536 >>>> lxc-start 40.087 WARN lxc_cgfs - >>>> cgfs.c:lxc_cgroup_get_container_info:1100 - Not attaching to cgroup cpuset >>>> unknown to /containers mycnt >>>> lxc-start 40.087 DEBUG lxc_start - >>>> start.c:setup_signal_fd:259 - sigchild handler set >>>> lxc-start 40.089 DEBUG lxc_console - >>>> console.c:lxc_console_peer_default:500 - opening /dev/tty for console peer >>>> lxc-start 40.089 DEBUG lxc_console - >>>> console.c:lxc_console_peer_default:506 - using '/dev/tty' as console >>>> lxc-start 40.089 DEBUG lxc_console - >>>> console.c:lxc_console_sigwinch_init:179 - 509 got SIGWINCH fd 9 >>>> lxc-start 40.089 DEBUG lxc_console - >>>> console.c:lxc_console_winsz:88 - set winsz dstfd:6 cols:0 rows:0 >>>> lxc-start 40.089 INFO lxc_start - start.c:lxc_init:451 - >>>> 'mycnt' is initialized >>>> lxc-start 40.095 DEBUG lxc_start - >>>> start.c:__lxc_start:1137 - Not dropping cap_sys_boot or watching utmp >>>> lxc-start 40.097 INFO lxc_start - >>>> start.c:resolve_clone_flags:848 - Cloning a new user namespace >>>> lxc-start 40.112 DEBUG lxc_conf - >>>> conf.c:instantiate_veth:2703 - instantiated veth 'veth-mycnt/vethNPPNHE', >>>> index is '12' >>>> lxc-start 40.112 INFO lxc_cgroup - >>>> cgroup.c:cgroup_init:65 - cgroup driver cgroupfs initing for mycnt >>>> lxc-start 40.122 DEBUG lxc_conf - >>>> conf.c:lxc_assign_network:3120 - move 'eth0' to '512' >>>> lxc-start 40.122 NOTICE lxc_start - start.c:do_start:667 - >>>> switching to gid/uid 0 in new user namespace >>>> lxc-start 40.125 DEBUG lxc_conf - >>>> conf.c:setup_rootfs:1284 - mounted '/containers/mycnt/rootfs' on >>>> '/usr/lib/lxc/rootfs' >>>> lxc-start 40.125 INFO lxc_conf - >>>> conf.c:setup_utsname:919 - 'mycnt' hostname has been setup >>>> lxc-start 40.153 DEBUG lxc_conf - >>>> conf.c:setup_netdev:2479 - 'eth0' has been setup >>>> lxc-start 40.153 INFO lxc_conf - >>>> conf.c:setup_network:2500 - network has been setup >>>> lxc-start 40.153 INFO lxc_conf - >>>> conf.c:mount_autodev:1148 - Mounting /dev under /usr/lib/lxc/rootfs >>>> lxc-start 40.153 INFO lxc_conf - >>>> conf.c:mount_autodev:1169 - Mounted tmpfs onto /usr/lib/lxc/rootfs/dev >>>> lxc-start 40.153 INFO lxc_conf - >>>> conf.c:mount_autodev:1187 - Mounted /dev under /usr/lib/lxc/rootfs >>>> lxc-start 40.153 DEBUG lxc_conf - conf.c:mount_entry:1735 >>>> - remounting /shared on /usr/lib/lxc/rootfs/tmp to respect bind or remount >>>> optios >>>> lxc-start 40.153 DEBUG lxc_conf - conf.c:mount_entry:1750 >>>> - (at remount) flags for /shared was 0, required extra flags are 0 >>>> lxc-start 40.153 DEBUG lxc_conf - conf.c:mount_entry:1759 >>>> - mountflags already was 4098, skipping remount >>>> lxc-start 40.153 DEBUG lxc_conf - conf.c:mount_entry:1785 >>>> - mounted '/shared' on '/usr/lib/lxc/rootfs/tmp', type 'none' >>>> lxc-start 40.153 INFO lxc_conf - >>>> conf.c:mount_file_entries:2034 - mount points have been setup >>>> lxc-start 40.153 INFO lxc_conf - >>>> conf.c:fill_autodev:1215 - Creating initial consoles under >>>> /usr/lib/lxc/rootfs/dev >>>> lxc-start 40.153 INFO lxc_conf - >>>> conf.c:fill_autodev:1226 - Populating /dev under /usr/lib/lxc/rootfs >>>> lxc-start 40.154 INFO lxc_conf - >>>> conf.c:fill_autodev:1258 - Populated /dev under /usr/lib/lxc/rootfs >>>> lxc-start 40.154 INFO lxc_conf - >>>> conf.c:setup_dev_console:1515 - console has been setup >>>> lxc-start 40.154 INFO lxc_conf - >>>> conf.c:do_tmp_proc_mount:3576 - I am 1, /proc/self points to '' >>>> lxc-start 40.161 INFO lxc_conf - >>>> conf.c:do_tmp_proc_mount:3602 - Mounted /proc in container for security >>>> transition >>>> lxc-start 40.500 ERROR lxc_conf - >>>> conf.c:prepare_ramfs_root:1378 - Invalid argument - Unable to umount /proc >>>> lxc-start 40.500 ERROR lxc_conf - conf.c:lxc_setup:3917 - >>>> failed to set rootfs for 'mycnt' >>>> lxc-start 40.500 ERROR lxc_start - start.c:do_start:699 - >>>> failed to setup the container >>>> lxc-start 40.500 ERROR lxc_sync - sync.c:__sync_wait:51 - >>>> invalid sequence number 1. expected 2 >>>> lxc-start 40.500 WARN lxc_conf - >>>> conf.c:lxc_delete_network:2996 - failed to remove interface 'eth0' >>>> lxc-start 40.515 ERROR lxc_start - >>>> start.c:__lxc_start:1164 - failed to spawn 'mycnt' >>>> lxc-start 40.669 ERROR lxc_start_ui - >>>> lxc_start.c:main:344 - The container failed to start. >>>> lxc-start 40.669 ERROR lxc_start_ui - >>>> lxc_start.c:main:348 - Additional information can be obtained by setting >>>> the --logfile and --logprio. >>>> >>>> >>>> On 07/28/2015 05:09 PM, Serge Hallyn wrote: >>>>> Quoting Przemyslaw Rudy ([email protected]): >>>>>> Referring to this patch: >>>>>> https://lists.linuxcontainers.org/pipermail/lxc-devel/2014-October/010604.html >>>>>> >>>>>> Starting lxc with userns in prepare_ramfs_root() I got -1 from: >>>>> >>>>> rv or errno? >>>>> >>>>>> if (umount2("./proc", MNT_DETACH)) { >>>>>> >>>>>> Shall this error be rather ignored in case of userns? Thus the same >>>>>> logic as for other mount points processed by mentioned function? >>>>> >>>>> Can you give full (-l trace -o debug.output) startup log? >>>>> _______________________________________________ >>>>> lxc-devel mailing list >>>>> [email protected] >>>>> http://lists.linuxcontainers.org/listinfo/lxc-devel >>>>> >>>> _______________________________________________ >>>> lxc-devel mailing list >>>> [email protected] >>>> http://lists.linuxcontainers.org/listinfo/lxc-devel >>> _______________________________________________ >>> lxc-devel mailing list >>> [email protected] >>> http://lists.linuxcontainers.org/listinfo/lxc-devel >>> >> _______________________________________________ >> lxc-devel mailing list >> [email protected] >> http://lists.linuxcontainers.org/listinfo/lxc-devel > _______________________________________________ > lxc-devel mailing list > [email protected] > http://lists.linuxcontainers.org/listinfo/lxc-devel > _______________________________________________ lxc-devel mailing list [email protected] http://lists.linuxcontainers.org/listinfo/lxc-devel
