Re: block.db linking to 2 disks

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Fixed, seems even though the block.db/block.wal had correct perms the disk entry under /dev was missing ceph:ceph ownership after the reboot for some reason.

Sorry for adding extra emails to your mailbox, but hopefully this may help someone else one day.

On Mon, Feb 25, 2019 at 11:09 PM Ashley Merrick <singapore@xxxxxxxxxxxxxx> wrote:
Sorry with log level 20 turned on for bluestore / bluefs

   -31> 2019-02-25 15:07:27.842 7f2bfbd71240 10 bluestore(/var/lib/ceph/osd/ceph-8) _open_db initializing bluefs
   -30> 2019-02-25 15:07:27.842 7f2bfbd71240 10 bluefs add_block_device bdev 1 path /var/lib/ceph/osd/ceph-8/block.db
   -29> 2019-02-25 15:07:27.842 7f2bfbd71240  1 bdev create path /var/lib/ceph/osd/ceph-8/block.db type kernel
   -28> 2019-02-25 15:07:27.842 7f2bfbd71240  1 bdev(0x5651277e6a80 /var/lib/ceph/osd/ceph-8/block.db) open path /var/lib/ceph/osd/ceph-8/block.db
   -27> 2019-02-25 15:07:27.842 7f2bfbd71240  1 bdev(0x5651277e6a80 /var/lib/ceph/osd/ceph-8/block.db) open size 107374182400 (0x1900000000, 100 GiB) block_size 4096 (4 KiB) rotational
   -26> 2019-02-25 15:07:27.842 7f2bfbd71240  1 bluefs add_block_device bdev 1 path /var/lib/ceph/osd/ceph-8/block.db size 100 GiB
   -25> 2019-02-25 15:07:27.842 7f2bfbd71240 10 bluestore(/var/lib/ceph/osd/ceph-8/block.db) _read_bdev_label
   -24> 2019-02-25 15:07:27.842 7f2bfbd71240 10 bluestore(/var/lib/ceph/osd/ceph-8/block.db) _read_bdev_label got bdev(osd_uuid 77703c4b-eb1d-4fae-a0e8-d6a80e55cd6e, size 0x1900000000, btime 2018-11-10 10:10:3                            9.072862, desc bluefs db, 0 meta)
   -23> 2019-02-25 15:07:27.842 7f2bfbd71240 10 bluefs add_block_device bdev 2 path /var/lib/ceph/osd/ceph-8/block
   -22> 2019-02-25 15:07:27.842 7f2bfbd71240  1 bdev create path /var/lib/ceph/osd/ceph-8/block type kernel
   -21> 2019-02-25 15:07:27.842 7f2bfbd71240  1 bdev(0x5651277e6e00 /var/lib/ceph/osd/ceph-8/block) open path /var/lib/ceph/osd/ceph-8/block
   -20> 2019-02-25 15:07:27.846 7f2bfbd71240  1 bdev(0x5651277e6e00 /var/lib/ceph/osd/ceph-8/block) open size 9834397171712 (0x8f1bfc00000, 8.9 TiB) block_size 4096 (4 KiB) rotational
   -19> 2019-02-25 15:07:27.846 7f2bfbd71240  1 bluefs add_block_device bdev 2 path /var/lib/ceph/osd/ceph-8/block size 8.9 TiB
   -18> 2019-02-25 15:07:27.846 7f2bfbd71240  1 bluefs mount
   -17> 2019-02-25 15:07:27.846 7f2bfbd71240 10 bluefs _open_super
   -16> 2019-02-25 15:07:27.846 7f2bfbd71240 10 bluefs _open_super superblock 54
   -15> 2019-02-25 15:07:27.846 7f2bfbd71240 10 bluefs _open_super log_fnode file(ino 1 size 0x100000 mtime 0.000000 bdev 0 allocated 500000 extents [1:0x1062700000+100000,0:0xf500000+400000])
   -14> 2019-02-25 15:07:27.846 7f2bfbd71240 20 bluefs _init_alloc
   -13> 2019-02-25 15:07:27.846 7f2bfbd71240 10 bluefs _replay
   -12> 2019-02-25 15:07:27.846 7f2bfbd71240 10 bluefs _replay log_fnode file(ino 1 size 0x100000 mtime 0.000000 bdev 0 allocated 500000 extents [1:0x1062700000+100000,0:0xf500000+400000])
   -11> 2019-02-25 15:07:27.846 7f2bfbd71240 10 bluefs _read h 0x565127411c80 0x0~1000 from file(ino 1 size 0x100000 mtime 0.000000 bdev 0 allocated 500000 extents [1:0x1062700000+100000,0:0xf500000+400000])
   -10> 2019-02-25 15:07:27.846 7f2bfbd71240 20 bluefs _read fetching 0x0~100000 of 1:0x1062700000+100000
    -9> 2019-02-25 15:07:27.862 7f2bfbd71240 20 bluefs _read left 0x100000 len 0x1000
    -8> 2019-02-25 15:07:27.862 7f2bfbd71240 20 bluefs _read got 4096
    -7> 2019-02-25 15:07:27.862 7f2bfbd71240 20 bluefs _replay need 0x4000 more bytes
    -6> 2019-02-25 15:07:27.862 7f2bfbd71240 10 bluefs _read h 0x565127411c80 0x1000~4000 from file(ino 1 size 0x100000 mtime 0.000000 bdev 0 allocated 500000 extents [1:0x1062700000+100000,0:0xf500000+400000]                            )
    -5> 2019-02-25 15:07:27.862 7f2bfbd71240 20 bluefs _read left 0xff000 len 0x4000
    -4> 2019-02-25 15:07:27.862 7f2bfbd71240 20 bluefs _read got 16384
    -3> 2019-02-25 15:07:27.862 7f2bfbd71240 10 bluefs _replay 0x0: txn(seq 1 len 0x4586 crc 0xfb7afd17)
    -2> 2019-02-25 15:07:27.862 7f2bfbd71240 20 bluefs _replay 0x0:  op_init
    -1> 2019-02-25 15:07:27.862 7f2bfbd71240 20 bluefs _replay 0x0:  op_alloc_add  0:0x1000~1ffff000
     0> 2019-02-25 15:07:27.866 7f2bfbd71240 -1 *** Caught signal (Segmentation fault) **
 in thread 7f2bfbd71240 thread_name:ceph-osd


On Mon, Feb 25, 2019 at 11:06 PM Ashley Merrick <singapore@xxxxxxxxxxxxxx> wrote:
So I was able to change the perms using :  chown -h ceph:ceph /var/lib/ceph/osd/ceph-6/block.db

However now I get the following when starting the OSD which then causes it to crash

 bluefs add_block_device bdev 2 path /var/lib/ceph/osd/ceph-8/block size 8.9 TiB
    -1> 2019-02-25 15:03:51.990 7f26d4777240  1 bluefs mount
     0> 2019-02-25 15:03:52.006 7f26d4777240 -1 *** Caught signal (Segmentation fault) **
 in thread 7f26d4777240 thread_name:ceph-osd

 ceph version 13.2.4 (b10be4d44915a4d78a8e06aa31919e74927b142e) mimic (stable)
 1: (()+0x9414c0) [0x55a8fc0b54c0]
 2: (()+0x12dd0) [0x7f26d5da8dd0]
 3: (BlueFS::_replay(bool, bool)+0x11ce) [0x55a8fc079e6e]
 4: (BlueFS::mount()+0xff) [0x55a8fc07d16f]
 5: (BlueStore::_open_db(bool, bool)+0x81c) [0x55a8fbfa9c3c]
 6: (BlueStore::_mount(bool, bool)+0x1a3) [0x55a8fbfd04a3]
 7: (OSD::init()+0x27d) [0x55a8fbbc250d]
 8: (main()+0x30a2) [0x55a8fba9ceb2]
 9: (__libc_start_main()+0xeb) [0x7f26d567a09b]
 10: (_start()+0x2a) [0x55a8fbb685aa]
 NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to interpret this.

Not sure if this just means the bluefs is corrupt or something I can try and repair.


On Mon, Feb 25, 2019 at 10:15 AM Ashley Merrick <singapore@xxxxxxxxxxxxxx> wrote:
After a reboot of a node I have one particular OSD that won't boot. (Latest Mimic)

When I "/var/lib/ceph/osd/ceph-8 # ls -lsh"

I get "   0 lrwxrwxrwx 1 root root 19 Feb 25 02:09 block.db -> '/dev/sda5 /dev/sdc5'"

For some reasons it is trying to link block.db to two disks, if I remove the block.db link and manually create the correct link the OSD still fails to start due to the perms on block.db file being root:root.

If I run a chmod it just goes back to root:root and the following shows in the OSD logs

2019-02-25 02:03:21.738 7f574b2a1240 -1 bluestore(/var/lib/ceph/osd/ceph-8) _open_db /var/lib/ceph/osd/ceph-8/block.db symlink exists but target unusable: (13) Permission denied
2019-02-25 02:03:21.738 7f574b2a1240  1 bdev(0x55dbf0a56700 /var/lib/ceph/osd/ceph-8/block) close
2019-02-25 02:03:22.034 7f574b2a1240 -1 osd.8 0 OSD:init: unable to mount object store
2019-02-25 02:03:22.034 7f574b2a1240 -1  ** ERROR: osd init failed: (13) Permission denied

Thanks
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

[Index of Archives]     [Information on CEPH]     [Linux Filesystem Development]     [Ceph Development]     [Ceph Large]     [Ceph Dev]     [Linux USB Development]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [xfs]


  Powered by Linux