Re: zfs: allocating allocated segment

Michael Gusek schrieb:
> Peter Schuller schrieb:
>>> I'm running a File server with zfs, 64 Bit, 4 GB Ram and RELENG_7. 
>>> Two days ago i uploaded a file via ftp to the server and the server 
>>> is crashing. After reboot FreeBSD can't import the zfs-pool. There 
>>> is a kernel-message:
>>>
>>> panic: Solaris(panic): zfs: allocating allocated 
>>> segment(offset=123456... size=74)
>>>
>>> cpuid = 0
>>> Uptime: 14m22s
>>> panic: bufwrite: buffer is not busy???
>>>
>>> Now i try to import the zfs-pool with a recent 8.0-current but with 
>>> the same result. It's very important to me to access the pool, so 
>>> did you have some idea's ?
>>>     
>>
>> The ZFS panic is from space_map.c in space_map_add(), happening via
>> zfs_panic_recover(). It in turn is affected by zfs_recover:
>>
>> /*                                                                                                                                                                                                                                             
>> * zfs_recover can be set to nonzero to attempt to recover 
>> from                                                                                                                                                                                
>> * otherwise-fatal errors, typically caused by on-disk corruption.  
>> When                                                                                                                                                                       
>> * set, calls to zfs_panic_recover() will turn into warning 
>> messages.                                                                                                                                                                          
>> */
>>
>> Setting the vfs.zfs.recover loader variable to 1 might possibly
>> help. However I have never tried using that option and I'm not
>> familiar with the code, so I have no idea how safe it is. In
>> particular since you then seem to be getting a secondary panic (the
>> "buffer is not busy" which is from ffs_vfsops.
>>
>> On another note:
>>
>> http://www.google.com/search?client=opera&rls=en&q=zfs:+allocating+allocated+segment&sourceid=opera&ie=utf-8&oe=utf-8 
>>
>>
>> indicates you're not the only person who has seen similar
>> errors. Unfortunately I cannot offer any insight other than to suggest
>> digging through the google results.
>>
>> Was the original crash, prior to the mount problem, purely a software
>> crash or was there, for example, a power outtage? I'm wondering
>> whether there is any particular reason to believe there was some
>> hardware/firmware fault causing corruption.
>>
>>   
> Thank you for your response Petter,
>
> on Monday i will give vfs.zfs.recover a try. The original crash was 
> beeing a ftp upload. There was'nt a power outtage or another kernel 
> message, so i don't think it is an hardware issue.
>
> Regards,
>
> Michael
So now i'm trying set vfs.zfs.recover to 1. But the kernel crashed with

panic: solaris assert: sm->sm_space == space (0xxfceee400 == 0xfceefc00) 
file:
/usr/src/sys/modules/zfs/../../cddl/contrib/opensolaris/uts/common/fs/zfs/space_map.c, 
line: 357

I't seems the zfs structure is corrupt ? And then, is there a chance to 
repair this ?

Greetings,

Michael