zfs discuss - Aug 2010 - Postmortem - file system recovered

I am writing to follow-up my post regarding a file system that became
inaccessible despite a clean pool:
http://opensolaris.org/jive/thread.jspa?messageID=494651

Several people helped but Victor Latushkin was instrumental in diagnosing/fixing
the issue.

Background:
The file system became inaccessible shortly after I began using NexentaStor
3.03.  I had been executing several recursive chown/chgrp/chmod commands. 
I''ve test the computer several times with Memtest and have never had
any detectable hardware issues.

The symptom:
A ls command yielded a strange line for this file system.  It would not cd into
the directory/filesystem
"""?--------- ? ? ? ? ? myfilesystemname"""

The problem:
""" it has mode bits set that indicated that it is FIFO,
character device and directory at the same time"""

The fix:
"""the trick was to modify mode in in-kernel buffer containing
znode_phys_t and then force ZFS to flush it out to disk."""

Outstanding questions:
1)  Is there a bug in ZFS or NexentaStor that resulted in the mode bits being
set incorrectly?
2)  When the mode bits were set to an invalid state why did ZFS react
ambiguously instead of reporting a clear error?  Why not report this when
encountered or at least have a tool to scrub the file system (not the pool)
looking for invalid data.  Without Victor''s help I would have never in
100 years discovered what the issue was.
3)  Could this error be recovered from automatically?   This was the root of a
zfs file system and regardless of the mode bits it was probably clear that it
should be treated as a directory.

Thanks for everyone''s help with diagnosing this.

-brian
-- 
This message posted from opensolaris.org

0n Sun, Aug 29, 2010 at 08:09:22PM -0700, Brian wrote: 

    >The fix:
    >"""the trick was to modify mode in in-kernel buffer
containing znode_phys_t and then force ZFS to flush it out to
disk."""

Can you give an example of how you did this ?

   -Alex

IMPORTANT: This email remains the property of the Department of Defence and is
subject to the jurisdiction of section 70 of the Crimes Act 1914. If you have
received this email in error, you are requested to contact the sender and delete
the email.

I am afraid I can''t describe the exact procedure that eventually fixed
the file system as I merely observed it while Victor was logged into my system. 
I am quoting from the explanation he provided but if he reads this perhaps he
could add whatever details seem pertinent.
-- 
This message posted from opensolaris.org