[sheepdog-users] Sheepdog disk usage

Hitoshi Mitake mitake.hitoshi at gmail.com
Thu May 18 06:20:15 CEST 2017


On Tue, May 9, 2017 at 10:09 PM, Teun Kloosterman
<teunkloosterman at gmail.com> wrote:
> Hi all,
>
> I am still experiencing this problem which is blocking me from using
> sheepdog in a testing environment, let alone production. Right now, for me,
> the 'dog vdi rollback' command does not delete any changes made after the
> snapshot, it just makes them inaccessible.
>
> Is there anyone here using the rollback feature who had any success in
> freeing up the disk space?
> Or is it my workflow and understanding of snapshot/rollback features? Should
> I use clone/delete instead?

Which version of sheepdog are you using (0.8.x?) ? The fine grained GC
is only available in 0.9.x or newer. Could you try newer releases?

Thanks,
Hitoshi

>
> I have enabled the discard option on the host and slave machines. I haven't
> seen any difference in using different virtualized controllers, I have
> tried: VirtIO, SCSI and SATA.
>
> Regards,
>   Teun
>
>
> PS. Some command output:
>
> root at host:~# dog vdi rollback -s 1 debian
> This operation dicards any changes made since the previous
> snapshot was taken.  Continue? [yes/no]: yes
> root at host:~# dog vdi rollback -s 2 centos
> This operation dicards any changes made since the previous
> snapshot was taken.  Continue? [yes/no]: yes
> root at host:~# dog node info
> Id    Size    Used    Avail    Use%
>  0    84 GB    25 GB    59 GB     30%
>  1    84 GB    28 GB    56 GB     33%
>  2    84 GB    25 GB    60 GB     29%
> Total    253 GB    79 GB    175 GB     31%
>
> Total virtual image size    40 GB
>
> --- Ansible Run
>
> root at host:~# dog node info
> Id    Size    Used    Avail    Use%
>  0    84 GB    27 GB    57 GB     32%
>  1    84 GB    31 GB    54 GB     36%
>  2    84 GB    27 GB    58 GB     31%
> Total    253 GB    85 GB    168 GB     33%
>
> Total virtual image size    40 GB
> root at host:~# dog vdi rollback -s 2 centos
> This operation dicards any changes made since the previous
> snapshot was taken.  Continue? [yes/no]: yes
> root at host:~# dog vdi rollback -s 1 debian
> This operation dicards any changes made since the previous
> snapshot was taken.  Continue? [yes/no]: yes
> root at host:~# dog node info
> Id    Size    Used    Avail    Use%
>  0    84 GB    27 GB    57 GB     32%
>  1    84 GB    31 GB    54 GB     36%
>  2    84 GB    27 GB    58 GB     31%
> Total    253 GB    85 GB    168 GB     33%
>
> Total virtual image size    40 GB
>
> --- Ansible Run
>
> root at host:~# dog node info
> Id    Size    Used    Avail    Use%
>  0    84 GB    33 GB    52 GB     38%
>  1    84 GB    37 GB    48 GB     43%
>  2    84 GB    32 GB    52 GB     37%
> Total    253 GB    102 GB    152 GB     40%
>
> Total virtual image size    40 GB
> root at host:~# dog vdi list
>   Name        Id    Size    Used  Shared    Creation time   VDI id  Copies
> Tag
> s centos       2   20 GB  204 MB  1.4 GB 2017-04-05 17:57   8d7d4a      2
>   centos       0   20 GB  1.9 GB  700 MB 2017-04-18 11:02   8d7d57      2
> s debian       1   20 GB  2.4 GB  0.0 MB 2017-04-05 17:12   b1f0b0      2
>   debian       0   20 GB  2.4 GB  1.3 GB 2017-04-18 11:02   b1f0bd      2
> root at host:~# dog vdi rollback -s 2 centos
> This operation dicards any changes made since the previous
> snapshot was taken.  Continue? [yes/no]: yes
> root at host:~# dog vdi rollback -s 1 debian
> This operation dicards any changes made since the previous
> snapshot was taken.  Continue? [yes/no]: yes
> root at host:~# dog node info
> Id    Size    Used    Avail    Use%
>  0    84 GB    33 GB    52 GB     38%
>  1    84 GB    37 GB    48 GB     43%
>  2    84 GB    32 GB    52 GB     37%
> Total    253 GB    102 GB    152 GB     40%
>
> Total virtual image size    40 GB
>
> --- Ansible Run
>
> root at host:~# dog vdi rollback -s 2 centos
> This operation dicards any changes made since the previous
> snapshot was taken.  Continue? [yes/no]: yes
> root at host:~# dog vdi rollback -s 1 debian
> This operation dicards any changes made since the previous
> snapshot was taken.  Continue? [yes/no]: yes
> root at host:~# dog node info
> Id    Size    Used    Avail    Use%
>  0    84 GB    36 GB    48 GB     42%
>  1    84 GB    40 GB    44 GB     47%
>  2    84 GB    35 GB    49 GB     41%
> Total    253 GB    112 GB    142 GB     44%
>
> Total virtual image size    40 GB
>
>
>
> On 22 March 2017 at 11:29, Teun Kloosterman <teunkloosterman at gmail.com>
> wrote:
>>
>> Hi,
>>
>> I'm afraid the discard option did not help with my issue.
>>
>> Is there any problem with the rollback command?
>> I use it extensively for resetting these test machines to vanilla state.
>>
>> Regards,
>>   Teun
>>
>> ---
>> root at host:~# dog node info
>> Id    Size    Used    Avail    Use%
>>  0    94 GB    88 GB    6.3 GB     93%
>>  1    99 GB    99 GB    0.0 MB    100%
>>  2    94 GB    86 GB    7.6 GB     91%
>> Total    286 GB    272 GB    14 GB     95%
>>
>> Total virtual image size    40 GB
>>
>> root at host:~# dog vdi list
>>   Name        Id    Size    Used  Shared    Creation time   VDI id  Copies
>> Tag
>> s centos       1   20 GB  2.2 GB  0.0 MB 2017-02-02 11:41   8d7d49      2
>> s pxe          1   20 GB  3.5 GB  0.0 MB 2017-02-02 11:41   917450      2
>> s debian       2   20 GB  920 MB  1.4 GB 2017-03-15 11:23   b1f0b6      2
>>   debian       0   20 GB  0.0 MB  2.2 GB 2017-03-22 10:42   b1f0ca      2
>>   arch         0   20 GB  1.9 GB  0.0 MB 2017-02-08 10:40   b23369      2
>>
>> root at host:~# dog vdi rollback -s 2 debian
>> This operation dicards any changes made since the previous
>> snapshot was taken.  Continue? [yes/no]: yes
>> Failed to create VDI debian: Failed to write to requested VDI
>>
>> On 31 January 2017 at 17:18, Vasiliy Tolstov <v.tolstov at selfip.ru> wrote:
>>>
>>> You must enable discard inside VM, enable discard in qemu, and don't use
>>> fully preallocated images
>>>
>>> 31 Янв 2017 г. 18:41 пользователь "Teun Kloosterman"
>>> <teunkloosterman at gmail.com> написал:
>>>
>>> Hi all,
>>>
>>> I've installed a sheepdog cluster for testing purposes on some desktop
>>> PCs. These have 120GB SSDs, which is not big, I know, but should suffice.
>>> They all run stock Debian Jessie with stable sheepdog 0.8.3-2 installed.
>>>
>>> Now I'm running into the issue that sheepdog consumes all disk space on
>>> these machines, all the way down to zero, and I cannot help it. These
>>> machines use less than 5GB for themselves and more than 270GB on sheepdog
>>> data. The images should consume (1.5 + 2.3) * 2 = 7.5GB or a maximum of 20 *
>>> 3 * 2 = 120 G. All data is located in the /mnt/sheep/0 data folder and the
>>> .stale folder is empty.
>>>
>>> Can anyone explain to me what this data is and how I can manage it?
>>>
>>> root at host03:/# dog node info
>>> Id    Size    Used    Avail    Use%
>>>  0    89 GB    89 GB    0.0 MB    100%
>>>  1    97 GB    91 GB    5.5 GB     94%
>>>  2    97 GB    92 GB    4.6 GB     95%
>>> Total    282 GB    272 GB    10 GB     96%
>>>
>>> Total virtual image size    20 GB
>>>
>>> root at host03:/# dog vdi list
>>>   Name        Id    Size    Used  Shared    Creation time   VDI id
>>> Copies  Tag
>>> s centos       1   20 GB  1.5 GB  0.0 MB 2016-09-14 13:12   8d7d6a      2
>>>   pxe          0   20 GB  0.0 MB  0.0 MB 2016-10-24 17:18   917450      2
>>> s debian       1   20 GB  2.3 GB  0.0 MB 2016-09-14 13:12   b1f0d2      2
>>>
>>> Kind regards,
>>>   Teun Kloosterman
>>>
>>> --
>>> sheepdog-users mailing lists
>>> sheepdog-users at lists.wpkg.org
>>> https://lists.wpkg.org/mailman/listinfo/sheepdog-users
>>>
>>>
>>
>
>
> --
> sheepdog-users mailing lists
> sheepdog-users at lists.wpkg.org
> https://lists.wpkg.org/mailman/listinfo/sheepdog-users
>


More information about the sheepdog-users mailing list