Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 18 Dec 2014 12:46:30 -0500
From:      Adam McDougall <mcdouga9@egr.msu.edu>
To:        freebsd-stable@freebsd.org
Subject:   Re: geom_multipath and zfs doesn't work
Message-ID:  <549312F6.1070007@egr.msu.edu>
In-Reply-To: <5493090D.2030406@fsn.hu>
References:  <5493090D.2030406@fsn.hu>

next in thread | previous in thread | raw e-mail | index | archive | help
On 12/18/2014 12:04, Nagy, Attila wrote:
> Hi,
> 
> Running stable/10@r273159 on FC disks (through isp) I can't create a zfs
> pool.
> 
> What I have is a simple device, accessible on /dev/multipath/sas0,
> backed by da0 and da4:
> # gmultipath status
>            Name   Status  Components
>  multipath/sas0  OPTIMAL  da0 (ACTIVE)
>                           da4 (READ)
> 
> When I issue a
> zpool create data /dev/multipath/sas0
> command, zpool starts to eat 100% CPU:
> # procstat -k 3924
>   PID    TID COMM             TDNAME KSTACK
>  3924 100128 zpool            - <running>
> 
> gstat shows that there is one uncompleted read IO on multipath/sas0 and
> the queue length constantly grows on da0:
> # gstat -b
> dT: 1.030s  w: 1.000s
>  L(q)  ops/s    r/s   kBps   ms/r    w/s   kBps   ms/w %busy Name
>  124402    146      0      0    0.0      0      0    0.0 100.5  da0
> I can use these devices finely with dd.
> 
> What's going on here?

I have a hunch.  Try sysctl vfs.zfs.vdev.trim_on_init=0 before zpool create?



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?549312F6.1070007>