Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
sysfs Problem
View unanswered posts
View posts from last 24 hours
View posts from last 7 days

 
Reply to topic    Gentoo Forums Forum Index Deutsches Forum (German)
View previous topic :: View next topic  
Author Message
Raze
n00b
n00b


Joined: 13 Jul 2009
Posts: 59
Location: Germany

PostPosted: Fri Jan 16, 2015 11:48 am    Post subject: sysfs Problem Reply with quote

Hallo zusammen,

folgendes Problem bei OpenVZ-Container nach einem Kernel-Update (*) stellt sich mir gerade:

Code:
/etc/init.d/net.eth0 start
mkdir: das Verzeichnis „/sys/fs/cgroup/openrc“ kann nicht angelegt werden: Die Datei existiert bereits
mount: permission denied
mkdir: das Verzeichnis „/sys/fs/cgroup/cpuset“ kann nicht angelegt werden: Die Datei existiert bereits
mount: special device cpuset does not exist
mkdir: das Verzeichnis „/sys/fs/cgroup/cpu“ kann nicht angelegt werden: Die Datei existiert bereits
mount: special device cpu does not exist
mkdir: das Verzeichnis „/sys/fs/cgroup/cpuacct“ kann nicht angelegt werden: Die Datei existiert bereits
mount: special device cpuacct does not exist
mkdir: das Verzeichnis „/sys/fs/cgroup/devices“ kann nicht angelegt werden: Die Datei existiert bereits
mount: special device devices does not exist
mkdir: das Verzeichnis „/sys/fs/cgroup/freezer“ kann nicht angelegt werden: Die Datei existiert bereits
mount: special device freezer does not exist
mkdir: das Verzeichnis „/sys/fs/cgroup/net_cls“ kann nicht angelegt werden: Die Datei existiert bereits
mount: special device net_cls does not exist
mkdir: das Verzeichnis „/sys/fs/cgroup/blkio“ kann nicht angelegt werden: Die Datei existiert bereits
mount: special device blkio does not exist
mkdir: das Verzeichnis „/sys/fs/cgroup/perf_event“ kann nicht angelegt werden: Die Datei existiert bereits
mount: special device perf_event does not exist
mkdir: das Verzeichnis „/sys/fs/cgroup/net_prio“ kann nicht angelegt werden: Die Datei existiert bereits
mount: special device net_prio does not exist
 * ERROR: sysfs failed to start
 * ERROR: cannot start net.eth0 as sysfs would not start


(*) = aktuell 2.6.32-openvz-042stab102.9-amd64, vormals 2.6.32-openvz-042stab094.8-amd64

Das Problem existiert erst seit der neusten Kernel-Version. Ich kann mir darauf nicht wirklich einen Reim machen, das diff zwischen den .configs halte ich für unbedenklich:

Code:
diff -w config-2.6.32-openvz-042stab094.8-amd64 config-2.6.32-openvz-042stab102.9-amd64
3,4c3,4
< # Linux kernel version: 2.6.32-openvz-042stab094.8-amd64
< # Tue Dec 16 20:35:46 2014
---
> # Linux kernel version: 2.6.32-openvz-042stab102.9-amd64
> # Fri Dec 19 20:33:32 2014
195a196,197
> CONFIG_HAVE_PERF_REGS=y
> CONFIG_HAVE_PERF_USER_STACK_DUMP=y
459c461
< CONFIG_ACPI_IPMI=y
---
> CONFIG_ACPI_IPMI=m
479a482
> CONFIG_ACPI_EXTLOG=m
548a552
> # CONFIG_PCI_REALLOC_ENABLE_AUTO is not set
1052a1057
> CONFIG_NET_RX_BUSY_POLL=y
1451a1457
> # CONFIG_SCSI_MVSAS_TASKLET is not set
1595a1602
> CONFIG_BLK_DEV_DM_BUILTIN=y
1604a1612,1615
> CONFIG_DM_CACHE=m
> CONFIG_DM_CACHE_MQ=m
> CONFIG_DM_CACHE_CLEANER=m
> CONFIG_DM_ERA=m
1796a1808,1811
> CONFIG_I40E=m
> # CONFIG_I40E_VXLAN is not set
> # CONFIG_I40E_DCB is not set
> CONFIG_I40EVF=m
1807a1823,1824
> CONFIG_MLX5_CORE=m
> CONFIG_MLX5_DEBUG=y
1812a1830,1831
> CONFIG_QLCNIC_DCB=y
> CONFIG_QLCNIC_HWMON=y
2343a2363
> CONFIG_HYPERV_KEYBOARD=m
2409c2429
< CONFIG_IPMI_HANDLER=y
---
> CONFIG_IPMI_HANDLER=m
2412c2432
< CONFIG_IPMI_SI=y
---
> CONFIG_IPMI_SI=m
2571a2592,2593
> CONFIG_POWERCAP=y
> CONFIG_INTEL_RAPL=m
2663a2686
> CONFIG_INTEL_POWERCLAMP=m
2753a2777
> CONFIG_MFD_RTSX_PCI=m
3168a3193
> CONFIG_DRM_KMS_FB_HELPER=y
3176a3202
> # CONFIG_DRM_I2C_NXP_TDA998X is not set
3187a3214,3216
> CONFIG_DRM_I915_FBDEV=y
> # CONFIG_DRM_I915_PRELIMINARY_HW_SUPPORT is not set
> CONFIG_DRM_I915_UMS=y
3500a3530
> CONFIG_HID_MULTITOUCH=m
3547d3576
< # CONFIG_USB_XHCI_HCD_DEBUGGING is not set
3726a3756
> CONFIG_MMC_REALTEK_PCI=m
3741a3772
> CONFIG_MEMSTICK_REALTEK_PCI=m
3780a3812
> CONFIG_INFINIBAND_QIB_DCA=y
3785a3818
> CONFIG_MLX5_INFINIBAND=m
3788c3821,3822
< # CONFIG_INFINIBAND_OCRDMA is not set
---
> CONFIG_INFINIBAND_OCRDMA=m
> # CONFIG_INFINIBAND_USNIC is not set
4015a4050
> CONFIG_UEFI_CPER=y
4327a4363
> # CONFIG_DEBUG_WW_MUTEX_SLOWPATH is not set
4418a4455
> CONFIG_EARLY_PRINTK_EFI=y
4518a4556,4557
> CONFIG_CRYPTO_CRCT10DIF=y
> CONFIG_CRYPTO_CRCT10DIF_PCLMUL=y
4529c4568
< CONFIG_CRYPTO_SHA256=m
---
> CONFIG_CRYPTO_SHA256=y
4569a4609,4613
> CONFIG_CRYPTO_DRBG_MENU=m
> CONFIG_CRYPTO_DRBG_HMAC=y
> CONFIG_CRYPTO_DRBG_HASH=y
> CONFIG_CRYPTO_DRBG_CTR=y
> CONFIG_CRYPTO_DRBG=m
4624d4667
< CONFIG_LIST_SORT=y


Patchnote: http://openvz.org/Download/kernel/rhel6/042stab102.9

Code:
Since 042stab094.8:

    Rebase to new RHEL6 kernel 2.6.32-504.1.3.el6 (security and bug fixes, see links below)
    netfilter: xt_connlimit: netns support (#3061)
    proc: reject ve context change of perms of node's proc files (PSBM-27933)
    cpt: typo fix in restore_posix_lock() (PSBM-30178)
    cpt/ipc: unbalanced preempt count in sysvipc_walk_sem (#3125, PSBM-30252)
    cpt: ipc sem fixup changes (#3125)
    ve/net: add a separate field for NETIF_F_VIRTUAL and NETIF_F_VENET (PSBM-29226)
    cpt: get rid of explicit i_size manipulation (PSBM-28962)
    ms/ext4: lock i_mutex when truncating orphan inodes
    ms/net: tcp: use correct net ns in cookie_v4_check() (PSBM-29553)
    rst: verbose debug for rst_restore_process() (#3085, PSBM-25446)
    cpt: put ct after nf_conntrack_hash_check_insert (PSBM-29589)
    cpt: Restore file locks owners (PSBM-29240)
    cpt: Dump whole siginfo_t::_sifields (PSBM-29266)
    cpt: Allow to receive pending data from unix socketpair with closed second end (PSBM-29264)
    ve/sysfs: create /sys/block/devName & /sys/dev/block/MAJ:MIN for CT ploops (PSBM-29112)
    ve/net: venet_set_op() cleanup
    cpt: Correct dump tmpfs having child mounts (PSBM-29265)
    fuse: introduce fuse_release_ff()
    fuse: fix deadlock in fuse_flush() (PSBM-29381)
    fuse: fix erroneous unlock_page() in fuse_send_writepages()
    ext4: update defragmentation codebase
    ext4: cleanup GFP flags inside resize path
    ve/net/ppp: fixed oops in oops in ppp_register_net_channel() (#3114, PSBM-29975)
    nfs: fixed nfs_fattr/nfs_fh leak in nfs_lookup_revalidate() (PSBM-29924)
    cpt: remove unused CPT_TEST_CAPS ioctl code
    cpt: fix TEST_VECAPS ioctl wrt xsave (#3012)
    cpt: Restore deleted delayed files in /tmp directory (PSBM-29582)
    rst: Zeroify ITIMER_PROF and ITIMER_VIRTUAL counting errors (PSBM-29857)
    ext4: fsync/mfsync cleanup
    mm: debug memory allocation causing fs re-entrance
    mm: compaction: restore irq only if lruvec lock is released (PSBM-29961)
    arch: x86: assure cpuinfo inside ct is subset of host's one (PSBM-30688)
    cpt: Fix wrong restore of conntrack
    cpt: unbalanced cpt_get/release_buf (#3139, PSBM-30546)
    x86/xsaves: Detect xsaves/xrstors feature
    x86, cpufeature: Rename X86_FEATURE_DTS to X86_FEATURE_DTHERM (PSBM-30577)
    arch: x86: check if cpuid level exists while reporting cpuinfo
    arch: x86: report cpuid:0000000d_1 in cpuinfo (PSBM-30577)
    arch: x86: only enable cpuid faulting for containers (PSBM-30516)
    arch: x86: disable cpuid faulting on reboot and cpu offline (PSBM-30548)
    cpt: missing i_mutex in find_linked_dentry() (#3145)
    typo in restore_one_vfsmount (#3041)
    cpt: Delete debug print in fixup_lock_pid()
    cpt: Check for retval of nf_conntrack_hash_check_insert() in undump_one_ct() (PSBM-30598)
    cgroup: do not inherit CGRP_SELF_DESTRUCTION flag
    netfilter: af_info: add network namespace parameter to route hook
    netfilter: nf_conntrack_h323: lookup route from proper net namespace
    xt_TCPMSS: lookup route from proper net namespace
    ub: account Shmem to meminfo:Cached (PSBM-28682)
    cgroup: Show /proc/cgroups inside container (PSBM-29815)
    xt_connbytes: enable CT accounting in proper net namespace
    cpt: dump packet socket mclist (PSBM-27254)
    cpt: fix ns pid in dump_one_flock (PSBM-30473)
    cpt: Fix restore of deleted hardlinks (PSBM-30318)
    cgroup: cpt -- Dont BUG_ON if something failed (PSBM-29815)
    cpt: dump_content_fifo fixup lockdep complain (PSBM-30266)
    cpt: cpt/restore continued/stopped flags of struct signal_struct (PSBM-29351)
    mm: add sysctl vm.force_scan_thresh (PSBM-29968)
    "cpu faulting" support for "cpu pools" feature
    VDSO/UTS virtualization for openSUSE 13.2 support
    fs/mnt: don't release unhold namespace_sem
    net/rtnetlink: rtnl_getlink() - allow query info by ifname (PSBM-30315)
    vedev: allow different symlinks source names for ve_device symlink (PSBM-30024)
    vedev: use correct ve context kobj when create symlinks for ve_device (PSBM-30024)
    vzcpt: Do not call nf_nat_cleanup_conntrack() in undump_one_ct() (#3151, PSBM-30844)


Weiterhin scheint es als wäre /sys problemlos gemountet:

Code:
 mount
/var/lib/vz/private/112 on / type simfs (rw,relatime)
proc on /proc type proc (rw,relatime)
sysfs on /sys type sysfs (rw,relatime)
tmpfs on /run type tmpfs (rw,nosuid,nodev,relatime,mode=755)
udev on /dev type devtmpfs (rw,nosuid,relatime,size=10240k,mode=755)
devpts on /dev/pts type devpts (rw,nosuid,noexec,relatime,gid=5,mode=620,ptmxmode=000)
shm on /dev/shm type tmpfs (rw,nosuid,nodev,noexec,relatime)
cgroup_root on /sys/fs/cgroup type tmpfs (rw,nosuid,nodev,noexec,relatime,size=10240k,mode=755)
/var/lib/vz/private/112 on /tmp type simfs (rw,noexec,relatime)
binfmt_misc on /proc/sys/fs/binfmt_misc type binfmt_misc (rw,nosuid,nodev,noexec,relatime)


Hat jemand einen Ansatz für mich den ich verfolgen könnte?
Back to top
View user's profile Send private message
Raze
n00b
n00b


Joined: 13 Jul 2009
Posts: 59
Location: Germany

PostPosted: Fri Jan 16, 2015 1:17 pm    Post subject: Reply with quote

Ich konnte das Problem soeben lösen.

Auf Systemen, auf denen sysfs ein eigenes init-Script hat ( bin mal gerade zu faul heraus zu suchen ab welcher openrc-Version das der Fall ist ) muss in der rc.conf folgendes gesetzt sein:

Code:
rc_controller_cgroups="NO"


Anschließend funktioniert wieder alles wie gehabt.

#Problemediedieweltnichtbraucht :roll:

Edit:

Es scheint doch nicht zweifelsfrei anhand dessen festlegbar sein ob sysfs ein separates init-Script hat. An der Behebung des Problems ändert sich aber nichts.
Back to top
View user's profile Send private message
Raze
n00b
n00b


Joined: 13 Jul 2009
Posts: 59
Location: Germany

PostPosted: Tue Jan 20, 2015 9:49 am    Post subject: Reply with quote

Code:
Calculating dependencies... done!
[ebuild     U  ] sys-apps/busybox-1.23.0 [1.21.0] USE="ipv6 pam static -debug% -livecd -make-symlinks -math -mdev -savedconfig (-selinux) -sep-usr -syslog -systemd" 2201 KiB
[ebuild     U  ] sys-apps/openrc-0.13.8 [0.12.4] USE="ncurses netifrc pam unicode -debug -newnet (-prefix) (-selinux) -static-libs -tools" 148 KiB
[ebuild     U  ] sys-apps/kmod-19 [18-r1] USE="tools zlib -debug -doc -lzma -python -static-libs" PYTHON_TARGETS="python2_7 python3_3 -python3_4" 1440 KiB
[ebuild     U  ] net-fs/nfs-utils-1.3.1-r1 [1.2.9-r3] USE="ipv6 libmount nfsidmap nfsv4 tcpd uuid -caps -kerberos -nfsdcld -nfsv41 (-selinux)" 768 KiB
[ebuild     U  ] sys-fs/udev-init-scripts-27 [26-r2] 4 KiB
[blocks b      ] <sys-fs/udev-init-scripts-27 ("<sys-fs/udev-init-scripts-27" is blocking sys-apps/openrc-0.13.8)


Nach den Updates von openrc und den udev-init-scripts fährt mein portage-server auch ohne die Anpassung in der rc.conf sauber hoch, sysfs wird korrekt eingehangen, Netzwerkschnittstellen fahren sauber hoch.
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Deutsches Forum (German) All times are GMT
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum