Removing log device from ZFS pool

[Anthony Ananich]

Hi, Steve,

Thanks for the quick reply. So if I got you right, the data in the
pool is lost, right?

This doc assert the log device is not critical:
http://docs.oracle.com/cd/E19253-01/819-5461/ghbxs/

Disks die from time to time. That's a reality. However in this
particular case I'm nearly sure that all the transactions was
committed to the persistent disks and ZIL was empty just due to
specifics of the scenario.

The only problem is that I'm using zfs v5000 and can not use Oracle
Solaris to restore the pool. It does not support v5000.

Do I have to patch sources to walk around this problem? Is there a
more easy way?

Thanks,
Anthony





On Tue, May 20, 2014 at 12:58 PM, Steven Hartland
<killing at multiplay.co.uk> wrote:
> Simply don't that will break the world, log devices must be persistent
>
>    Regards
>    Steve
> ----- Original Message ----- From: "Anthony Ananich"
> <anton.ananich at inpun.com>
> To: <zfs-devel at freebsd.org>
> Sent: Tuesday, May 20, 2014 10:46 AM
> Subject: Removing log device from ZFS pool
>
>
>> Hi!
>>
>> Here is what I tried to do:
>>
>> 1) create zfs pool (two hard disks)
>> 2) add log device to the pool
>> 3) add cache device to the pool
>> 4) reboot server
>>
>> In this scenario log device dies during the reboot.
>>
>> -----
>> # zpool list tank
>> NAME      SIZE  ALLOC   FREE    CAP  DEDUP  HEALTH  ALTROOT
>> tank      928G   274G   654G    29%  1.00x  ONLINE  -
>>
>> # zpool status tank
>>  pool: tank
>> state: ONLINE
>>  scan: none requested
>> config:
>> NAME                    STATE     READ WRITE CKSUM
>> tank                    ONLINE       0     0     0
>>  mirror-0              ONLINE       0     0     0
>>    gpt/disk1           ONLINE       0     0     0
>>    gpt/disk2           ONLINE       0     0     0
>> errors: No known data errors
>>
>> # mdconfig -a -t swap -s 128m -u 1
>> # zpool add tank log /dev/md1
>> # zpool status tank
>>  pool: tank
>> state: ONLINE
>>  scan: none requested
>> config:
>> NAME                    STATE     READ WRITE CKSUM
>> raptor2                 ONLINE       0     0     0
>>  mirror-0              ONLINE       0     0     0
>>    gpt/disk1           ONLINE       0     0     0
>>    gpt/disk2           ONLINE       0     0     0
>> logs
>>  md1                   ONLINE       0     0     0
>> errors: No known data errors
>> -----
>>
>> As long as I'm using volatile device /dev/md1 in this example, it is
>> destroyed during reboot.
>>
>> Due to documentation this is not critical, I can just ignore unsaved
>> data and discard uncommitted log entries.
>>
>> However it does not work for me in reality:
>>
>> -----
>> # zpool status tank
>> pool: tank
>> state: FAULTED
>> status: An intent log record could not be read.
>> Waiting for adminstrator intervention to fix the faulted pool.
>> action: Either restore the affected device(s) and run 'zpool online',
>> or ignore the intent log records by running 'zpool clear'.
>>  see: http://illumos.org/msg/ZFS-8000-K4
>> scan: none requested
>> config:
>>
>> NAME                    STATE     READ WRITE CKSUM
>> tank                    FAULTED      0     0     0
>> mirror-0               ONLINE       0     0     0
>>   gpt/disk1            ONLINE       0     0     0
>>   gpt/disk2            ONLINE       0     0     0
>> logs
>> 6324139563861643487   UNAVAIL      0     0     0  was /dev/md1
>>
>> # zpool clear tank
>> cannot clear errors for tank: one or more devices is currently unavailable
>>
>> # zpool remove tank 6324139563861643487
>> cannot open 'tank': pool is unavailable
>>
>> # zpool online tank md1
>> cannot open ‘tank’: pool is unavailable
>>
>> -----
>>
>> O wonder if I'm doing something wrong and this is expected behaviour?
>> Or that's just a bug?
>>
>> I'm using zfs v5000 at FreeBSD 9.2-RELEASE.
>>
>> Regards,
>> Anthony
>> _______________________________________________
>> zfs-devel at freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/zfs-devel
>> To unsubscribe, send any mail to "zfs-devel-unsubscribe at freebsd.org"
>
>