Re: [zfs-discuss] space_map again nuked!!

2010-09-22 Thread Orvar Korvar
Did you see this thread?
http://opensolaris.org/jive/thread.jspa?messageID=500659#500659

He had problems with ZFS. It turned out to be faulty RAM. ZFS is so sensitive 
it detects and reports problems to you. No other filesystem does that, so you 
think ZFS is problematic and switch. But the other filesystems is slowly 
corrupting your data, because they dont notice problems.

In other words, I suspect the problem is not in ZFS, but somewhere else. Maybe 
hardware? Maybe BIOS?
-- 
This message posted from opensolaris.org
___
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss


Re: [zfs-discuss] space_map again nuked!!

2010-09-22 Thread Stephan Ferraro
 Did you see this thread?
 http://opensolaris.org/jive/thread.jspa?messageID=5006
 59#500659

I get on this link:

Tomcat http error 500:
The server encountered an internal error () that prevented it from fulfilling 
this request.

 
 He had problems with ZFS. It turned out to be faulty
 RAM. ZFS is so sensitive it detects and reports
 problems to you. No other filesystem does that, so
 you think ZFS is problematic and switch. But the
 other filesystems is slowly corrupting your data,
 because they dont notice problems.
 
 In other words, I suspect the problem is not in ZFS,
 but somewhere else. Maybe hardware? Maybe BIOS?

That could be possible because its very cheap hardware from the hosting service 
Hetzner.
I will run a memtest to get more information about it.
If I detect errors on the ram and they will exchange it, how can I repair my 
defective spacemap?
-- 
This message posted from opensolaris.org
___
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss


Re: [zfs-discuss] space_map again nuked!!

2010-09-22 Thread Victor Latushkin

On Sep 23, 2010, at 1:11 AM, Stephan Ferraro wrote:
 
 He had problems with ZFS. It turned out to be faulty
 RAM. ZFS is so sensitive it detects and reports
 problems to you. No other filesystem does that, so
 you think ZFS is problematic and switch. But the
 other filesystems is slowly corrupting your data,
 because they dont notice problems.
 
 In other words, I suspect the problem is not in ZFS,
 but somewhere else. Maybe hardware? Maybe BIOS?
 
 That could be possible because its very cheap hardware from the hosting 
 service Hetzner.

As far as I understand it have been working for you for some time. How long was 
that?

Have you been doing something new and different with your system lately?

 I will run a memtest to get more information about it.
 If I detect errors on the ram and they will exchange it, how can I repair my 
 defective spacemap?

Here the same trick with aok/zfs_recover may work, but in general the only 
supported way at the moment is to backup all data, destroy and recreate pool 
and then restore data back.

Cheers
Victor
___
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss


Re: [zfs-discuss] space_map again nuked!!

2010-09-19 Thread Victor Latushkin
On Sep 18, 2010, at 11:37 AM, Stephan Ferraro wrote:

 I'm really angry against ZFS:

Emotions rarely help to get to the root cause...

 My server no more reboots because the ZFS spacemap is again corrupt.
 I just replaced the whole spacemap by recreating a new zpool from scratch and 
 copying back the data with zfs send  zfs receive.
 Did it copied corrupt spacemap?!

Definitely no. It suggests that you may have and issue with hardware, for 
example with memory, or CPU and/or cache, or some other components.

regards
Victor
___
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss


[zfs-discuss] space_map again nuked!!

2010-09-18 Thread Stephan Ferraro
I'm really angry against ZFS:
My server no more reboots because the ZFS spacemap is again corrupt.
I just replaced the whole spacemap by recreating a new zpool from scratch and 
copying back the data with zfs send  zfs receive.
Did it copied corrupt spacemap?!

For me its now terminated. I loss to much time and money with this experimental 
filesystem.
My version is Zpool  ZFS of snv_111b.

I will switch back to Linux to ext4.
-- 
This message posted from opensolaris.org
___
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss


Re: [zfs-discuss] space_map again nuked!!

2010-09-18 Thread Stephan Ferraro
For the developpers, this is my error message:
Sep 18 09:43:56 global genunix: [ID 361072 kern.warning] WARNING: zfs: freeing 
free segment (offset=379733483520 size=4096)
-- 
This message posted from opensolaris.org
___
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss