Uploaded image for project: 'Lustre'
  1. Lustre
  2. LU-20558

utils: tunefs.lustre unconditionally sets autodegrade=on on a ZFS OST

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Unresolved
    • Icon: Medium Medium
    • None
    • None
    • None
    • 3
    • 9223372036854775807

      On a ZFS OST, tunefs.lustre unconditionally appends "autodegrade=on" to the target parameters. As a result it silently discards an autodegrade value given on the command line.

       

      1. Calling mkfs.lustre sets autodegrade to on by default

      # truncate -s 2G /tmp/ost1.img
# zpool create -f -o cachefile=none lustre-ost1 /tmp/ost1.img
# mkfs.lustre --ost --backfstype=zfs --fsname=lustre --index=0 --mgsnode=xxx.xx.xx.xx@tcp lustre-ost1/ost1   Permanent disk data:
Device:     lustre-ost1/ost1
Target:     lustre:OST0000
Index:      0
Lustre FS:  lustre
Mount type: zfs
Flags:      0x62
              (OST first_time update )
Persistent mount opts:
Parameters: mgsnode=xxx.xx.xx.xx@tcp autodegrade=on
checking for existing Lustre data: not found
mkfs_cmd = zfs create -o canmount=off  lustre-ost1/ost1
  xattr=sa
  dnodesize=auto
  recordsize=1M
  compression=off
Writing lustre-ost1/ost1 properties
  lustre:mgsnode=xxx.xx.xx.xx@tcp
  lustre:autodegrade=on
  lustre:version=1
  lustre:flags=98
  lustre:index=0
  lustre:fsname=lustre
  lustre:svname=lustre:OST0000
# zfs get -H -o value lustre:autodegrade lustre-ost1/ost1
on

       

      2. Try to set autodegrade to off by tunefs.lustre but doesn't work.

      # tunefs.lustre --param autodegrade=off lustre-ost1/ost1
checking for existing Lustre data: found   Read previous values:
Device:     lustre-ost1/ost1
Target:     lustre-OST0000
Index:      0
Lustre FS:  lustre
Mount type: zfs
Flags:      0x62
              (OST first_time update )
Persistent mount opts:
Parameters: autodegrade=on mgsnode=xxx.xx.xx.xx@tcp   Permanent disk data:
Device:     lustre-ost1/ost1
Target:     lustre:OST0000
Index:      0
Lustre FS:  lustre
Mount type: zfs
Flags:      0x62
              (OST first_time update )
Persistent mount opts:
Parameters: mgsnode=xxx.xx.xx.xx@tcp autodegrade=off autodegrade=on
Writing lustre-ost1/ost1 properties
  lustre:mgsnode=xxx.xx.xx.xx@tcp
  lustre:autodegrade=off
  lustre:autodegrade=on
  lustre:version=1
  lustre:flags=98
  lustre:index=0
  lustre:fsname=lustre
  lustre:svname=lustre:OST0000
# zfs get -H -o value lustre:autodegrade lustre-ost1/ost1
on

       

      3. Try to erase autodegrade by tunefs.lustre but doesn't work either.

      # tunefs.lustre --erase-param autodegrade lustre-ost1/ost1
checking for existing Lustre data: found   Read previous values:
Device:     lustre-ost1/ost1
Target:     lustre-OST0000
Index:      0
Lustre FS:  lustre
Mount type: zfs
Flags:      0x62
              (OST first_time update )
Persistent mount opts:
Parameters: autodegrade=on mgsnode=xxx.xx.xx.xx@tcp   Permanent disk data:
Device:     lustre-ost1/ost1
Target:     lustre:OST0000
Index:      0
Lustre FS:  lustre
Mount type: zfs
Flags:      0x62
              (OST first_time update )
Persistent mount opts:
Parameters: mgsnode=xxx.xx.xx.xx@tcp autodegrade=on
Writing lustre-ost1/ost1 properties
  lustre:mgsnode=xxx.xx.xx.xx@tcp
  lustre:autodegrade=on
  lustre:version=1
  lustre:flags=98
  lustre:index=0
  lustre:fsname=lustre
  lustre:svname=lustre:OST0000
# zfs get -H -o value lustre:autodegrade lustre-ost1/ost1
on

            xyang Xiao Yang
            xyang Xiao Yang
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated: