Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 14 Jul 2015 11:19:12 -0700
From:      Chris Stankevitz <chris@stankevitz.com>
To:        Freddie Cash <fjwcash@gmail.com>
Cc:        FreeBSD Filesystems <freebsd-fs@freebsd.org>
Subject:   Re: Offline/online a drive from a mounted RW RAIDZx pool: resilver needed?
Message-ID:  <55A552A0.3010408@stankevitz.com>
In-Reply-To: <CAOjFWZ5V2DdKfsDcFrtrwOKNpx4=UDGEcWEj5%2BYpbYkHw05qZg@mail.gmail.com>
References:  <55A4461F.6020308@stankevitz.com>	<CAOjFWZ6vuxtCy01ro=oHKvZPpKi6BbaAhtWZk%2B6gDk9aPdTy%2Bw@mail.gmail.com>	<55A53BE2.7060609@stankevitz.com> <CAOjFWZ5V2DdKfsDcFrtrwOKNpx4=UDGEcWEj5%2BYpbYkHw05qZg@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 7/14/15 9:46 AM, Freddie Cash wrote:
> If nothing is written to the pool (or, at least that specific vdev)
> while the drive is removed, then the metadata on the drive will match
> the pool, and it can be onlined without a resilver.
>
> If only a little bit of data has been written, and a short bit of time
> has passed, ​
> ​then the drive will only need to sync (resilver) that changed data to
> the drive.
>
> If a lot of data has been written, or a long time has passed (not sure
> what qualifies as "long", maybe until the last txg synced to the disk is
> no longer in the txg history of the pool?), then the entire drive will
> be resilvered.
>
> Or, something along those lines.  That's what I've observed by doing
> various tests (offline drive, pull drive, write to pool, online drive;
>   or offline drive, replace with new one, start resilver, new drive
> dies, stick in old drive, online it, wait for resilver; and other
> scenarios).

Freddie,

Thank you.  That certainly agrees with what I experienced.  You can 
imagine my terror when I yanked the wrong drive... and my fear/confusion 
when the drive did not resilver as I expected.

Chris



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?55A552A0.3010408>