Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 21 Jan 2019 12:35:24 +0000
From:      Steve O'Hara-Smith <steve@sohara.org>
To:        Yasuhiro KIMURA <yasu@utahime.org>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: Performance degradation of ZFS?
Message-ID:  <20190121123524.3c97bdfb2254e3393a58b831@sohara.org>
In-Reply-To: <20190121.210941.2275299827563542964.yasu@utahime.org>
References:  <20190121.210941.2275299827563542964.yasu@utahime.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 21 Jan 2019 21:09:41 +0900 (JST)
Yasuhiro KIMURA <yasu@utahime.org> wrote:

> Is it possible that I/O performance degrade while using ZFS for a long
> time? And if answer is yes is there any way to recover performance
> without making full backup and restore?

	How is the pool configured ? How full is it ? How well is the data
striped (are some vdevs fuller than others) ? Are any of the drives showing
trouble ? Do you run a regular scrub ? What does zpool status show ?

	I can think of several ways in which ZFS performance can become
degraded (and there are probably a good many I haven't thought of) but time
passing is not enough alone, something has to be going astray. Those
questions should help pin down what.

	What it will take to recover depends greatly on the cause, you may
find (for example) that replacing a flaky drive (not failing just making too
many retries) will bring dramatic improvement or you might be unlucky and
have developed hideously imbalanced stripes for which there is no cure but
to wipe and rebuild.

-- 
Steve O'Hara-Smith <steve@sohara.org>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20190121123524.3c97bdfb2254e3393a58b831>