Check status of background destroy
Freddie Cash
fjwcash at gmail.com
Thu Feb 16 21:33:45 UTC 2017
On Fri, Jan 6, 2017 at 9:03 AM, Freddie Cash <fjwcash at gmail.com> wrote:
> On Wed, Jan 4, 2017 at 10:43 AM, Xin LI <delphij at gmail.com> wrote:
>
>> Try setting vfs.zfs.free_max_blocks to 100000 (this is an arbitrary
>> number, can be larger, limits maximum blocks being freed per txg; see
>> https://svnweb.freebsd.org/base?view=revision&revision=271532 for
>> details).
>>
>> You can inspect the progress with DTrace by looking at zfs-dbgmsg.
>>
>
> This is looking promising.
>
> Booting off an 11.0-based mfsBSD didn't work (locked up after "Trying to
> mount root from /dev/md0").
>
> Booting off a 10.3-based mfsBSD worked. Set that sysctl, then imported
> the pool without issues. The box has now been running for just over 24
> hours.
>
> ARC is sitting at 93 GB, with Wired at 116 GB. Previously, the ARC would
> only be around 10 GB with Wired over 120 GB. "zpool iostat" shows lots of
> reads and writes happening to the pool. And the Avail space is slowly
> ticking upwards.
>
> Now just need to wait and see how long it takes to free up all 40 TB of
> referenced space. :)
>
> Thanks for the info!
>
> Just to close out this thread, the 11.0-based mfsBSD LiveCD worked to
allow the background destroy to complete successfully. Afterward, I was
able to import the storage pool back into the 10.3 install running off the
root pool. This server is back online and running normally.
Thanks for all the help and pointers everyone!
--
Freddie Cash
fjwcash at gmail.com
More information about the freebsd-fs
mailing list