Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 5 Dec 2025 13:50:16 +0000
From:      void <void@f-m.fm>
To:        freebsd-fs@freebsd.org
Subject:   Re: zpool setup clarification
Message-ID:  <aTLjGLcInHAy5JAo@int21h>
In-Reply-To: <20251203081050.1a2880af@arc.aei.uni-hannover.de>
References:  <aSxnG-HKDYZ0L3lQ@int21h> <b4b75e5e-071c-446f-8298-25f41a609334@app.fastmail.com> <aS8TWJKWqSGeAmo3@int21h> <CANCZdfrxz-swKF5HN_8cZO1A4qmY_x1a1Su2BxAAOQ7JLo6EtQ@mail.gmail.com> <20251203081050.1a2880af@arc.aei.uni-hannover.de>

next in thread | previous in thread | raw e-mail | index | archive | help

On Wed, Dec 03, 2025 at 08:10:50AM +0100, Gerrit Kühn wrote:
>Am Tue, 2 Dec 2025 17:32:54 -0700
>schrieb Warner Losh <imp@bsdimp.com>:
>
>> I always 1MB align partitions, and put the first one with ZFS at a 1MB
>> boundary. Though some really large NVMe drives making this 16MB might not
>> be bad for performance.
>
>Same here:
>
>storage3:~ # gpart show da0
>=>         40  42970644400  da0  GPT  (20T)
>           40         2008       - free -  (1.0M)
>         2048  42970640384    1  freebsd-zfs  (20T)
>  42970642432         2008       - free -  (1.0M)
>
>This is still HDD, but I set up SSDs in pretty much the same way. Usually
>it is hard to get information on the internal magic (erase block size
>etc.) of SSDs. In the end, we usually struggle more from write patterns we
>often see here that only change a few bytes in many (several 10k) files
>every few minutes, leading to heavy write amplification.
>
>
>cu
>  Gerrit

I didn't see the other replies (after the initial one) until
after the pool was created. 

Still, *if* a partition needs to be created and therefore if the pool 
needs to be re-made, it's no problem. The disk is an external 
20tb seagate. There's no partition to align right now, I hope this has
sidestepped any possible write amplification issue. maybe not? 
zdb says it has an ashift of 12.

I deleted the partitions that were there (think it was exfat32) 
after removing what was on there by default and then created a (stripe)
pool for the disk.
-- 



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?aTLjGLcInHAy5JAo>