From mboxrd@z Thu Jan  1 00:00:00 1970
From: Yi Zhang <yizhan@redhat.com>
Subject: Re: lots of "md: export_rdev(sde)" printed after create IMSM RAID10
 with missing
Date: Mon, 12 Sep 2016 16:03:58 +0800
Message-ID: <a543dd7a-b582-98fa-3ba8-67500396c766@redhat.com>
References: <338941973.7699634.1473230038475.JavaMail.zimbra@redhat.com>
 <1648084319.7702644.1473230621059.JavaMail.zimbra@redhat.com>
 <20160908225607.GA66921@kernel.org>
 <7910bc85-f9c4-1ea3-76a6-40b819738537@intel.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 7bit
Return-path: <linux-raid-owner@vger.kernel.org>
In-Reply-To: <7910bc85-f9c4-1ea3-76a6-40b819738537@intel.com>
Sender: linux-raid-owner@vger.kernel.org
To: Artur Paszkiewicz <artur.paszkiewicz@intel.com>, Shaohua Li <shli@kernel.org>
Cc: linux-raid@vger.kernel.org, Jes.Sorensen@redhat.comJes.Sorensen@redhat.com
List-Id: linux-raid.ids



On 09/09/2016 08:56 PM, Artur Paszkiewicz wrote:
> On 09/09/2016 12:56 AM, Shaohua Li wrote:
>> On Wed, Sep 07, 2016 at 02:43:41AM -0400, Yi Zhang wrote:
>>> Hello
>>>
>>> I tried create one IMSM RAID10 with missing, found lots of "md: export_rdev(sde)" printed, anyone could help check it?
>>>
>>> Steps I used:
>>> mdadm -CR /dev/md0 /dev/sd[b-f] -n5 -e imsm
>>> mdadm -CR /dev/md/Volume0 -l10 -n4 /dev/sd[b-d] missing
>>>
>>> Version:
>>> 4.8.0-rc5
>>> mdadm - v3.4-84-gbd1fd72 - 25th August 2016
>> can't reproduce with old mdadm but can with upstream mdadm. Looks mdadm is
>> keeping write the new_dev sysfs entry.
>>
>> Jes, any idea?
>>
>> Thanks,
>> Shaohua
>>> Log:
>>> http://pastebin.com/FJJwvgg6
>>>
>>> <6>[  301.102007] md: bind<sdb>
>>> <6>[  301.102095] md: bind<sdc>
>>> <6>[  301.102159] md: bind<sdd>
>>> <6>[  301.102215] md: bind<sde>
>>> <6>[  301.102291] md: bind<sdf>
>>> <6>[  301.103010] ata3.00: Enabling discard_zeroes_data
>>> <6>[  311.714344] ata3.00: Enabling discard_zeroes_data
>>> <6>[  311.721866] md: bind<sdb>
>>> <6>[  311.721965] md: bind<sdc>
>>> <6>[  311.722029] md: bind<sdd>
>>> <5>[  311.733165] md/raid10:md127: not clean -- starting background reconstruction
>>> <6>[  311.733167] md/raid10:md127: active with 3 out of 4 devices
>>> <6>[  311.733186] md127: detected capacity change from 0 to 240060989440
>>> <6>[  311.774027] md: bind<sde>
>>> <6>[  311.810664] md: md127 switched to read-write mode.
>>> <6>[  311.819885] md: resync of RAID array md127
>>> <6>[  311.819886] md: minimum _guaranteed_  speed: 1000 KB/sec/disk.
>>> <6>[  311.819887] md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for resync.
>>> <6>[  311.819891] md: using 128k window, over a total of 234435328k.
>>> <6>[  316.606073] ata3.00: Enabling discard_zeroes_data
>>> <6>[  343.949845] capability: warning: `turbostat' uses 32-bit capabilities (legacy support in use)
>>> <6>[ 1482.314944] md: md127: resync done.
>>> <7>[ 1482.315086] RAID10 conf printout:
>>> <7>[ 1482.315087]  --- wd:3 rd:4
>>> <7>[ 1482.315089]  disk 0, wo:0, o:1, dev:sdb
>>> <7>[ 1482.315089]  disk 1, wo:0, o:1, dev:sdc
>>> <7>[ 1482.315090]  disk 2, wo:0, o:1, dev:sdd
>>> <7>[ 1482.315099] RAID10 conf printout:
>>> <7>[ 1482.315099]  --- wd:3 rd:4
>>> <7>[ 1482.315100]  disk 0, wo:0, o:1, dev:sdb
>>> <7>[ 1482.315100]  disk 1, wo:0, o:1, dev:sdc
>>> <7>[ 1482.315101]  disk 2, wo:0, o:1, dev:sdd
>>> <7>[ 1482.315101]  disk 3, wo:1, o:1, dev:sde
>>> <6>[ 1482.315220] md: recovery of RAID array md127
>>> <6>[ 1482.315221] md: minimum _guaranteed_  speed: 1000 KB/sec/disk.
>>> <6>[ 1482.315222] md: using maximum available idle IO bandwidth (but not more than 200000 KB/sec) for recovery.
>>> <6>[ 1482.315227] md: using 128k window, over a total of 117217664k.
>>> <6>[ 2697.184217] md: md127: recovery done.
>>> <7>[ 2697.524143] RAID10 conf printout:
>>> <7>[ 2697.524144]  --- wd:4 rd:4
>>> <7>[ 2697.524146]  disk 0, wo:0, o:1, dev:sdb
>>> <7>[ 2697.524146]  disk 1, wo:0, o:1, dev:sdc
>>> <7>[ 2697.524147]  disk 2, wo:0, o:1, dev:sdd
>>> <7>[ 2697.524148]  disk 3, wo:0, o:1, dev:sde
>>> <6>[ 2697.524632] md: export_rdev(sde)
>>> <6>[ 2697.549452] md: export_rdev(sde)
>>> <6>[ 2697.568763] md: export_rdev(sde)
>>> <6>[ 2697.587938] md: export_rdev(sde)
>>> <6>[ 2697.607271] md: export_rdev(sdeautomate)
>>> <6>[ 2697.626321] md: export_rdev(sdeautomateautomate)
>>> <6>[ 2697.645676] md: export_rdev(sde)
>>> <6>[ 2697.663211] md: export_rdev(sde)
>>> <6>[ 2697.681603] md: export_rdev(sde)
>>> <6>[ 2697.699117] md: export_rdev(sde)
>>> <6>[ 2697.716510] md: export_rdev(sde)
>>>
>>> Best Regards,
>>>    Yi Zhang
> Can you check if this fix works for you? If it does I'll send a proper
> patch for this.
Hello Artur
With your patch, no "md: export_rdev(sde)" printed after create raid10.

I found another problem, not sure whether it is reasonable, could you 
help confirm it, thanks.
When I create one container with 4 disks[1], and create one raid10 with 
3 disks(sd[b-d]) + 1 missing [2], but it finally bind the fourth disk: 
sde [3].

[1] mdadm -CR /dev/md0 /dev/sd[b-e] -n4 -e imsm
[2] mdadm -CR /dev/md/Volume0 -l10 -n4 /dev/sd[b-d] missing --size=500M
[3] # cat /proc/mdstat
Personalities : [raid10]
md127 : active raid10 sde[4] sdd[2] sdc[1] sdb[0]
       1024000 blocks super external:/md0/0 128K chunks 2 near-copies 
[4/4] [UUUU]

md0 : inactive sde[3](S) sdd[2](S) sdc[1](S) sdb[0](S)
       4420 blocks super external:imsm

unused devices: <none>
> Thanks,
> Artur
>
> diff --git a/super-intel.c b/super-intel.c
> index 92817e9..ffa71f6 100644
> --- a/super-intel.c
> +++ b/super-intel.c
> @@ -7789,6 +7789,9 @@ static struct mdinfo *imsm_activate_spare(struct active_array *a,
>   			IMSM_T_STATE_DEGRADED)
>   		return NULL;
>   
> +	if (get_imsm_map(dev, MAP_0)->map_state == IMSM_T_STATE_UNINITIALIZED)
> +		return NULL;
> +
>   	/*
>   	 * If there are any failed disks check state of the other volume.
>   	 * Block rebuild if the another one is failed until failed disks