Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 2 Jul 2020 10:40:50 +0300
From:      Andriy Gapon <avg@FreeBSD.org>
To:        Gordon Bergling <gbe@freebsd.org>, Allan Jude <allanjude@freebsd.org>
Cc:        freebsd-hackers@freebsd.org
Subject:   Re: Constant load of 1 on a recent 12-STABLE
Message-ID:  <0a10a722-3eec-456b-d771-97a49af89ad1@FreeBSD.org>
In-Reply-To: <20200623072332.GA54653@lion.0xfce3.net>
References:  <20200603101607.GA80381@lion.0xfce3.net> <c18664e8-b4e3-1402-48ed-3a02dc36ce29@freebsd.org> <20200603202929.GA65032@lion.0xfce3.net> <8b1498ea-e343-506e-79c7-c25b594808f0@freebsd.org> <20200604123720.GA63595@lion.0xfce3.net> <20200623072332.GA54653@lion.0xfce3.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On 23/06/2020 10:23, Gordon Bergling wrote:
> What is maybe also related to the problem is the following top output, 
> collected by 'top -HS'. The three zfskern threads generate little load, but that
> constant. The annoying part of the problem is, that the load of 1 is reported and
> that the hostsystem schedules the thread over all 4 physical cores and keeps the
> clockrate at the highest frequency.
> 
>   PID USERNAME    PRI NICE   SIZE    RES STATE    C   TIME    WCPU COMMAND
>    11 root        155 ki31     0B    64K CPU2     2  17:28  97.39% idle{idle: cpu2}
>    11 root        155 ki31     0B    64K CPU3     3  17:29  96.78% idle{idle: cpu3}
>    11 root        155 ki31     0B    64K CPU1     1  17:29  96.40% idle{idle: cpu1}
>    11 root        155 ki31     0B    64K RUN      0  17:25  96.13% idle{idle: cpu0}
>     8 root         -8    -     0B  1040K mmp->m   2   0:44   4.32% zfskern{mmp_thread_enter}
>     8 root         -8    -     0B  1040K mmp->m   1   0:44   4.28% zfskern{mmp_thread_enter}
>     8 root         -8    -     0B  1040K mmp->m   3   0:44   4.25% zfskern{mmp_thread_enter}
> 
> Does anyone has some insight where zfskern is spawned? I maybe can bisect then the resulting change
> which leads to this behaviour.

That's strange.  Those mmp threads shouldn't really be consuming much CPU.
Normally they should just sleep all the time.
Do you use the new ZFS multi-modifier protection feature?  I guess not.
Do you have 'multihost' property set on any of your pools?

The change you were looking for is
r361383 MFC r354804,r354806,r354807,r354885: 10499 Multi-modifier protection (MMP)

-- 
Andriy Gapon



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?0a10a722-3eec-456b-d771-97a49af89ad1>