[zfs-discuss] Re: Very poor write speeds

Gordan Bobic gordan.bobic at gmail.com
Mon Jan 6 11:17:00 EST 2014


Because scrub doesn't happen in LBA order.



On Mon, Jan 6, 2014 at 4:08 PM, Durval Menezes <durval.menezes at gmail.com>wrote:

> Hi Gordan,
>
> On Mon, Jan 6, 2014 at 1:10 PM, Gordan Bobic <gordan.bobic at gmail.com>wrote:
>
>> What does your "zpool list" say? How effective is your deduplication?
>>
>> As I mentioned before, dedup (if it is doing anything) will cause masive
>> data fragmentation which will deteriorate your pool performance toward at
>> most 12MB/s per vdev, assuming all your data ends up being written in the
>> maximum size 128KB blocks and your disks are 7200rpm.
>>
>> I don't think the performance you are seeing is not at all unusual - the
>> numbers seem in the right ballpark.
>>
>
> Humrmrmr... I've absolutely no experience with dedup (and in this case
> lack of experience seems to be definitely a blessing ;-)), but why would
> dedup impact scrub speed? AFAIK, dedup only runs through the available
> blocks in the pool and reports if any are unreadable and/or have checksum
> errors, and would not care if those blocks are deduped or not, no? Or would
> the dedup have to be "processed back" for the checksum to be verified?
>
> Cheers,
> --
>    Durval.
>
>
>
>>
>>
>>
>>
>> On Mon, Jan 6, 2014 at 3:01 PM, Ian Worthington <
>> ian.s.worthington at gmail.com> wrote:
>>
>>> Yesterday's scrub was the worst yet, almost 23 hours:
>>>
>>> 2014-01-05 03:00:01: starting scrub on zfspool
>>> 2014-01-06 01:48:22: scrub ended on zfspool
>>> 2014-01-06 01:48:22:   pool: zfspool
>>>  state: ONLINE
>>>   scan: scrub repaired 0 in 22h47m with 0 errors on Mon Jan  6 01:47:55 2014
>>> config:
>>>
>>> 	NAME                                          STATE     READ WRITE CKSUM
>>> 	zfspool                                       ONLINE       0     0     0
>>> 	  mirror-0                                    ONLINE       0     0     0
>>> 	    ata-WDC_WD10EFRX-68JCSN0_WD-WCC1U4257356  ONLINE       0     0     0
>>> 	    ata-WDC_WD10EFRX-68JCSN0_WD-WCC1U4299344  ONLINE       0     0     0
>>>
>>> errors: No known data errors
>>> 2014-01-06 01:48:22:  - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
>>> - - - - - - - - - -
>>>
>>>  To unsubscribe from this group and stop receiving emails from it, send
>>> an email to zfs-discuss+unsubscribe at zfsonlinux.org.
>>>
>>
>>  To unsubscribe from this group and stop receiving emails from it, send
>> an email to zfs-discuss+unsubscribe at zfsonlinux.org.
>>
>
>  To unsubscribe from this group and stop receiving emails from it, send an
> email to zfs-discuss+unsubscribe at zfsonlinux.org.
>

To unsubscribe from this group and stop receiving emails from it, send an email to zfs-discuss+unsubscribe at zfsonlinux.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.zfsonlinux.org/pipermail/zfs-discuss/attachments/20140106/73993777/attachment.html>


More information about the zfs-discuss mailing list