[zfs-auto-snapshot] Auto snapshots not deleted on volumes

Wolfram Tomalla w.tomalla at googlemail.com
Wed Apr 1 06:50:23 PDT 2009


Hi Niall, Hi Tim,

did you decide, on how this should be handeled? Or are the changes even done?

Greetings,

Wolfram

2009/3/24 Niall Power <Niall.Power at sun.com>:
> Hi Wolfram,
>
> I suspect this is a regression introduced by some recent enhancements
> to time slider.
>
> Time-slider deletes certain 0 sized snapshots in order to make life
> easier for the user  trying to find a specific file amongs a set of
> snapshots.
>
> This had a knock on effect of breaking the snapshot deletion
> functionality of zfs-auto-snapshot since it only destroys snapshots
> in recursive sets. So the recurisive chain could easily be broken
> by time-slider's actions.
>
> To fix this I implemented the snapshot cleanup mechanisms that
> zfs-auto-snapshot usually performs in time-slider (it has a cleanup
> service that monitors disk space, and also now deletes selected 0 sized
> snapshots).
>
> However time-slider does not consider volumes, only filesystems.
> This was a decision that was made in the interests of simplicity and
> user friendliness for most laptop workstation users. In your case
> the mixture of volumes and filesystems in the hierarchy under n-pool:
>
> npool/n-data/xVM/Vol (filesystem)
> npool/n-data/xVM/Vol/Win (volume)
>
> This has resulted in the snapshots of npool/n-data/xVM/Vol being deleted
> but no for npool/n-data/xVM/Vol/Win since that is a volume that time-slider
> ignores.
>
> There are 2 approaches to fixing this:
> - Make zfs-auto-snapshot more flexible about deletion so that it doesn't
>  break when the recursive chain is broken.
> - Enhance time-slider-cleanup to perform cleanup duties on volumes in
>  addition to just filesystems.
>
> Tim:
> Any thoughts on this? I think the better option would be to perform
> cleanup of volumes in time-slider-cleanup. The setup UI would continue
> to ignore volumes however.
>
> Thanks,
> Niall.
>
> Wolfram Tomalla wrote:
>>
>> Hi Niall,
>>
>> first I initially set up automatic snapshotting using Time Slider. But
>> when I finally found out that my mashine took about 15 minutes for a
>> reboot, because of these 2500 snapshots. I destroyed all volumes
>> including their snapshots, and also changed the settings for
>> com.sun:auto-snapshot manualy to be recursively inherited, and set it
>> manualy to true on on the npool filesystem.
>>
>> Greetings,
>>
>> Wolfram
>>
>> 2009/3/24 Niall Power <Niall.Power at sun.com>:
>>
>>>
>>> Hi Tom,
>>>
>>> Did you set up automatic snapshotting using Time Slider
>>> or did you enable zfs-auto-snapshot's SMF service yourself?
>>>
>>> Thanks,
>>> Niall.
>>>
>>> Wolfram Tomalla wrote:
>>>
>>>>
>>>> Hi!
>>>>
>>>>
>>>> Ich have a problem on Opensolaris snv_109 with zfs-auto-snapshots
>>>> enabled.
>>>> The automatical snapshots on volumes appear not to be deleted
>>>> automaticaly but on filesystems they are :
>>>>
>>>> wolfram at Stickstoff:~$ zfs list -r -t all npool/n-data/xVM/Vol
>>>> NAME   USED  AVAIL  REFER  MOUNTPOINT
>>>> npool/n-data/xVM/Vol       108K  2.98T  38.0K  /n-data/xVM/Vol
>>>> npool/n-data/xVM/Vol at zfs-auto-snap:daily-2009-03-24-00:00   0      -
>>>>  38.0K  -
>>>> npool/n-data/xVM/Vol at zfs-auto-snap:hourly-2009-03-24-14:00   0      -
>>>>  38.0K  -
>>>> npool/n-data/xVM/Vol at zfs-auto-snap:frequent-2009-03-24-14:15   0
>>>> -  38.0K  -
>>>> npool/n-data/xVM/Vol/Win        69.9K  2.98T  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-16:45
>>>> 30.0K      -  32.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:hourly-2009-03-22-17:00   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-17:00   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-17:15   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-17:30   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-17:45   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-18:00   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:hourly-2009-03-22-18:00   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-18:15   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-18:30   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-18:45   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-22-19:00   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:hourly-2009-03-22-19:00   0
>>>>  -  40.0K  -
>>>>
>>>> [..]
>>>>
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:hourly-2009-03-24-13:00    0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-24-13:00   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-24-13:15   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-24-13:30   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-24-13:45   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:hourly-2009-03-24-14:00   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-24-14:00   0
>>>>  -  40.0K  -
>>>> npool/n-data/xVM/Vol/Win at zfs-auto-snap:frequent-2009-03-24-14:15   0
>>>>  -  40.0K  -
>>>>
>>>>
>>>>
>>>> In this case npool/n-data/xVM/Vol is a filesystem that should host my
>>>> volumes and npool/n-data/xVM/Vol/Win is the only volume I have
>>>> currently.
>>>>
>>>> According to the settings I can not find anything that should prevent
>>>> the snapshots from beeing purged:
>>>>
>>>> wolfram at Stickstoff:~$ zfs get all npool/n-data/xVM/Vol
>>>> NAME                  PROPERTY               VALUE
>>>>  SOURCE
>>>> npool/n-data/xVM/Vol  type                   filesystem             -
>>>> npool/n-data/xVM/Vol  creation               Sun Mar 22 16:40 2009  -
>>>> npool/n-data/xVM/Vol  used                   108K                   -
>>>> npool/n-data/xVM/Vol  available              2.98T                  -
>>>> npool/n-data/xVM/Vol  referenced             38.0K                  -
>>>> npool/n-data/xVM/Vol  compressratio          1.00x                  -
>>>> npool/n-data/xVM/Vol  mounted                yes                    -
>>>> npool/n-data/xVM/Vol  quota                  none
>>>> default
>>>> npool/n-data/xVM/Vol  reservation            none
>>>> default
>>>> npool/n-data/xVM/Vol  recordsize             128K
>>>> default
>>>> npool/n-data/xVM/Vol  mountpoint             /n-data/xVM/Vol inherited
>>>> from npool/n-data
>>>> npool/n-data/xVM/Vol  sharenfs               rw,root=@192.168
>>>> inherited from npool/n-data
>>>> npool/n-data/xVM/Vol  checksum               on
>>>> default
>>>> npool/n-data/xVM/Vol  compression            off
>>>>  default
>>>> npool/n-data/xVM/Vol  atime                  off   inherited from npool
>>>> npool/n-data/xVM/Vol  devices                on
>>>> default
>>>> npool/n-data/xVM/Vol  exec                   on
>>>> default
>>>> npool/n-data/xVM/Vol  setuid                 on
>>>> default
>>>> npool/n-data/xVM/Vol  readonly               off
>>>>  default
>>>> npool/n-data/xVM/Vol  zoned                  off
>>>>  default
>>>> npool/n-data/xVM/Vol  snapdir                hidden
>>>> default
>>>> npool/n-data/xVM/Vol  aclmode                groupmask
>>>>  default
>>>> npool/n-data/xVM/Vol  aclinherit             restricted
>>>> default
>>>> npool/n-data/xVM/Vol  canmount               on
>>>> default
>>>> npool/n-data/xVM/Vol  shareiscsi             off    inherited from npool
>>>> npool/n-data/xVM/Vol  xattr                  on
>>>> default
>>>> npool/n-data/xVM/Vol  copies                 1
>>>>  default
>>>> npool/n-data/xVM/Vol  version                3                      -
>>>> npool/n-data/xVM/Vol  utf8only               off                    -
>>>> npool/n-data/xVM/Vol  normalization          none                   -
>>>> npool/n-data/xVM/Vol  casesensitivity        sensitive              -
>>>> npool/n-data/xVM/Vol  vscan                  off
>>>>  default
>>>> npool/n-data/xVM/Vol  nbmand                 off
>>>>  default
>>>> npool/n-data/xVM/Vol  sharesmb               off
>>>>  default
>>>> npool/n-data/xVM/Vol  refquota               none
>>>> default
>>>> npool/n-data/xVM/Vol  refreservation         none
>>>> default
>>>> npool/n-data/xVM/Vol  primarycache           all
>>>>  default
>>>> npool/n-data/xVM/Vol  secondarycache         all
>>>>  default
>>>> npool/n-data/xVM/Vol  usedbysnapshots        0                      -
>>>> npool/n-data/xVM/Vol  usedbydataset          38.0K                  -
>>>> npool/n-data/xVM/Vol  usedbychildren         69.9K                  -
>>>> npool/n-data/xVM/Vol  usedbyrefreservation   0                      -
>>>> npool/n-data/xVM/Vol  com.sun:auto-snapshot  true  inherited from npool
>>>>
>>>>
>>>>
>>>> wolfram at Stickstoff:~$ zfs get all npool/n-data/xVM/Vol/Win
>>>> NAME                      PROPERTY               VALUE
>>>>  SOURCE
>>>> npool/n-data/xVM/Vol/Win  type                   volume
>>>> -
>>>> npool/n-data/xVM/Vol/Win  creation               Sun Mar 22 16:42 2009
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  used                   69.9K
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  available              2.98T
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  referenced             40.0K
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  compressratio          1.00x
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  reservation            none
>>>> default
>>>> npool/n-data/xVM/Vol/Win  volsize                10G
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  volblocksize           8K
>>>> -
>>>> npool/n-data/xVM/Vol/Win  checksum               on
>>>> default
>>>> npool/n-data/xVM/Vol/Win  compression            off
>>>>  default
>>>> npool/n-data/xVM/Vol/Win  readonly               off
>>>>  default
>>>> npool/n-data/xVM/Vol/Win  shareiscsi             off   inherited from
>>>> npool
>>>> npool/n-data/xVM/Vol/Win  copies                 1
>>>>  default
>>>> npool/n-data/xVM/Vol/Win  refreservation         none
>>>> default
>>>> npool/n-data/xVM/Vol/Win  primarycache           all
>>>>  default
>>>> npool/n-data/xVM/Vol/Win  secondarycache         all
>>>>  default
>>>> npool/n-data/xVM/Vol/Win  usedbysnapshots        30.0K
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  usedbydataset          40.0K
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  usedbychildren         0
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  usedbyrefreservation   0
>>>>  -
>>>> npool/n-data/xVM/Vol/Win  com.sun:auto-snapshot  true   inherited from
>>>> npool
>>>>
>>>>
>>>> As a side effect of the volume snapshots not beeing deleted, my
>>>> mashine took awfully long for a reeboot after several days, because is
>>>> was busy creating the 5000 devices vor the 2500 snapshots.
>>>>
>>>> Do you have any idea, why the snapshots of the Volume are not deleted,
>>>> but the snapshots of the filesystems are, allthough all the settings
>>>> for the volume are inherited from the same source as the settings for
>>>> the Filesystem.
>>>>
>>>> Thanks for any info,
>>>>
>>>> Wolfram
>>>> _______________________________________________
>>>> zfs-auto-snapshot mailing list
>>>> zfs-auto-snapshot at opensolaris.org
>>>> http://mail.opensolaris.org/mailman/listinfo/zfs-auto-snapshot
>>>>
>>>>
>>>
>>>
>
>



More information about the zfs-auto-snapshot mailing list