Linux High IOWAIT updatedb can't be killed and crash with mdadm

high IO wait

  424 root      39  19  1900  848  552 D  0.0  0.0   0:00.91 updatedb

root       424  0.0  0.0   1900   848 ?        DN   Mar11   0:00 /usr/bin/updatedb -f sysfs?rootfs?bdev?proc?cpuset?binfmt_misc?debugfs?sockfs?usbfs?pipefs?anon_inodefs?futexfs?tmpfs?inotifyfs?eventpollfs?devpts?ramfs?hugetlbfs?mqueue?rpc_pipefs?autofs
root      2907  0.0  0.0   1768   492 ?        DN   Mar15   0:00 /usr/bin/updatedb -f sysfs?rootfs?bdev?proc?cpuset?binfmt_misc?debugfs?sockfs?usbfs?pipefs?anon_inodefs?futexfs?tmpfs?inotifyfs?eventpollfs?devpts?ramfs?hugetlbfs?mqueue?rpc_pipefs?autofs
root      8691  0.0  0.0   1768   492 ?        DN   Mar12   0:00 /usr/bin/updatedb -f sysfs?rootfs?bdev?proc?cpuset?binfmt_misc?debugfs?sockfs?usbfs?pipefs?anon_inodefs?futexfs?tmpfs?inotifyfs?eventpollfs?devpts?ramfs?hugetlbfs?mqueue?rpc_pipefs?autofs
root     10642  0.0  0.0   4004   696 pts/0    S+   00:38   0:00 grep updatedb
root     14937  0.0  0.0   1768   492 ?        DN   Mar16   0:00 /usr/bin/updatedb -f sysfs?rootfs?bdev?proc?cpuset?binfmt_misc?debugfs?sockfs?usbfs?pipefs?anon_inodefs?futexfs?tmpfs?inotifyfs?eventpollfs?devpts?ramfs?hugetlbfs?mqueue?rpc_pipefs?autofs
root     16697  0.0  0.0   1768   488 ?        DN   Mar13   0:00 /usr/bin/updatedb -f sysfs?rootfs?bdev?proc?cpuset?binfmt_misc?debugfs?sockfs?usbfs?pipefs?anon_inodefs?futexfs?tmpfs?inotifyfs?eventpollfs?devpts?ramfs?hugetlbfs?mqueue?rpc_pipefs?autofs
root     18790  0.0  0.0   1768   496 ?        DN   Mar18   0:00 /usr/bin/updatedb -f sysfs?rootfs?bdev?proc?cpuset?binfmt_misc?debugfs?sockfs?usbfs?pipefs?anon_inodefs?futexfs?tmpfs?inotifyfs?eventpollfs?devpts?ramfs?hugetlbfs?mqueue?rpc_pipefs?autofs
root     24783  0.0  0.0   1768   500 ?        DN   Mar14   0:00 /usr/bin/updatedb -f sysfs?rootfs?bdev?proc?cpuset?binfmt_misc?debugfs?sockfs?usbfs?pipefs?anon_inodefs?futexfs?tmpfs?inotifyfs?eventpollfs?devpts?ramfs?hugetlbfs?mqueue?rpc_pipefs?autofs
root     24922  0.0  0.0   1768   492 ?        DN   Mar17   0:00 /usr/bin/updatedb -f sysfs?rootfs?bdev?proc?cpuset?binfmt_misc?debugfs?sockfs?usbfs?pipefs?anon_inodefs?futexfs?tmpfs?inotifyfs?eventpollfs?devpts?ramfs?hugetlbfs?mqueue?rpc_pipefs?autofs
root     29655  0.0  0.0   1768   496 ?        DN   Mar19   0:00 /usr/bin/updatedb -f sysfs?rootfs?bdev?proc?cpuset?binfmt_misc?debugfs?sockfs?usbfs?pipefs?anon_inodefs?futexfs?tmpfs?inotifyfs?eventpollfs?devpts?ramfs?hugetlbfs?mqueue?rpc_pipefs?autofs


I noticed that updatedb was running for days and then this dmesg message:

(manually killing the process and also killall updatedb does not stop it or kill it)

INFO: task updatedb:424 blocked for more than 300 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
updatedb      D F79C8CF0  6476   424    419                     (NOTLB)
       e740bdd4 00000086 c30045c0 000e2832 f7acea4c 00000001 f79c8cf0 c30f8800
       000e2832 000f4240 556bd125 0024f991 f79c8e14 f7d14000 556bf321 0024f991
       00000002 00000000 00000001 00000000 c0419058 f7d14000 c06ecd80 f7a0cdec
 Call Trace:
 [<c0419058>] __wake_up+0x2a/0x3d
 [<c0625676>] io_schedule+0x3b/0x64
 [<c0460ead>] sync_page+0x0/0x3b
 [<c0460ee5>] sync_page+0x38/0x3b
 [<c062578e>] __wait_on_bit_lock+0x2a/0x52
 [<c0460e28>] __lock_page+0x52/0x59
 [<c0436000>] wake_bit_function+0x0/0x3c
 [<c04686c9>] truncate_inode_pages_range+0x20b/0x260
 [<f8c698d3>] journal_stop+0x208/0x213 [jbd]
 [<c0468727>] truncate_inode_pages+0x9/0xe
 [<f8ca5a1e>] ext3_delete_inode+0x13/0xba [ext3]
 [<f8ca5a0b>] ext3_delete_inode+0x0/0xba [ext3]
 [<c0499c51>] generic_delete_inode+0x91/0xfe
 [<c0499728>] iput+0x67/0x69
 [<c04970dd>] d_kill+0x19/0x32
 [<c049821d>] dput+0x19f/0x1ac
 [<c04909c3>] sys_renameat+0x15f/0x1af
 [<c04ebb8e>] _atomic_dec_and_lock+0x2a/0x44
 [<c0453cf8>] audit_syscall_entry+0x160/0x192
 [<c0490a24>] sys_rename+0x11/0x15
 [<c06289e7>] syscall_call+0x7/0xb
 =======================

kill -9 424
strace -p 424

Process 424 attached - interrupt to quit

attach: ptrace(PTRACE_ATTACH, ...): Operation not permitted

==========================
Eventually the system crashed with references to ext3 and inodes etc..
After a reboot it did the following:

md: md0: raid array is not clean -- starting background reconstruction
raid1: raid set md0 active with 2 out of 2 mirrors
md: ... autorun DONE.
md: syncing RAID array md0
md: Autodetecting RAID arrays.
md: autorun ...
md: ... autorun DONE.
md: minimum _guaranteed_ reconstruction speed: 1000 KB/sec/disc.
md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for reconstruction.
md: using 128k window, over a total of 511999424 blocks.
EXT3-fs: INFO: recovery required on readonly filesystem.
EXT3-fs: write access will be enabled during recovery.
kjournald starting.  Commit interval 5 seconds
EXT3-fs: md0: orphan cleanup on readonly fs
ext3_orphan_cleanup: deleting unreferenced inode 97467990
ext3_orphan_cleanup: deleting unreferenced inode 76153118
EXT3-fs: md0: 2 orphan inodes deleted
EXT3-fs: recovery complete.
EXT3-fs: mounted filesystem with ordered data mode.
 


Tags:

linux, iowait, updatedb, mdadmhigh, io, dn, mar, usr, bin, sysfs, rootfs, bdev, proc, cpuset, binfmt_misc, debugfs, sockfs, usbfs, pipefs, anon_inodefs, futexfs, tmpfs, inotifyfs, eventpollfs, devpts, ramfs, hugetlbfs, mqueue, rpc_pipefs, autofs, pts, grep, dmesg, manually, killall, info, task, blocked, quot, echo, sys, kernel, hung_task_timeout_secs, disables, cf, notlb, bdd, acea, bd, bf, ecd, cdec, __wake_up, io_schedule, ead, sync_page, ee, __wait_on_bit_lock, __lock_page, wake_bit_function, truncate_inode_pages_range, journal_stop, jbd, truncate_inode_pages, xe, ca, ext, _delete_inode, xba, generic_delete_inode, xfe, iput, dd, d_kill, dput, ac, sys_renameat, af, ebb, _atomic_dec_and_lock, audit_syscall_entry, sys_rename, syscall_call, xb, strace, attach, ptrace, ptrace_attach, permitted, references, inodes, etc, reboot, md, raid, array, reconstruction, active, mirrors, autorun, syncing, autodetecting, arrays, _guaranteed_, kb, disc, maximum, idle, bandwidth, fs, readonly, filesystem, enabled, kjournald, interval, orphan, cleanup, _orphan_cleanup, deleting, unreferenced, inode, deleted, mounted, mode,

Latest Articles

  • Cisco Unified Communication Manager (CUCM) - How To Add Phones
  • pptp / pptpd not working in DD-WRT iptables / router
  • systemd-journald high memory usage solution
  • How to Install FreePBX in Linux Debian Ubuntu Mint Guide
  • How To Install Cisco's CUCM (Cisco Unified Communication Manager) 12 Guide
  • Linux Ubuntu Redhat How To Extract Images from PDF
  • Linux and Windows Dual Boot Issue NIC Won't work After Booting Windows
  • Cisco CME How To Enable ACD hunt groups
  • How to install gns3 on Linux Ubuntu Mint
  • How to convert audio for Asterisk .wav format
  • Using Cisco CME Router with Asterisk as a dial-peer
  • Cisco CME How To Configure SIP Trunk VOIP
  • Virtualbox host Only Network Error Failed to save host network interface parameter - Cannot change gateway IP of host only network
  • Cisco CME and C7200 Router Testing and Learning Environment on Ubuntu 20+ Setup Tutorial Guide
  • Abusive IP ranges blacklist
  • How to Install Any OS on a Physical Drive from Windows Using VMware Workstation (Linux, Windows, BSD)
  • CDN Cloudflare how to set and preserve the real IP of the client without modifying application code on Apache
  • CentOS 7 fix Could not retrieve mirrorlist http://mirrorlist.centos.org/?release=7&arch=x86_64&repo=os&infra=container error was 14: curl#6 -
  • Ubuntu Debian How To Install Recommended Packages Automatically
  • How to set Linux Ubuntu Redhat Debian Command Line http https socks proxy for yum apt