What is actual status of SUJ in 9-STABLE?

Kirk McKusick mckusick at mckusick.com
Thu Mar 29 17:35:03 UTC 2012


> Date: Thu, 29 Mar 2012 12:13:44 +0400
> From: Lev Serebryakov <lev at freebsd.org>
> To: freebsd-fs at freebsd.org, freebsd-stable at freebsd.org
> Subject: What is actual status of SUJ in 9-STABLE?
> 
> Hello, Freebsd-fs.
> 
>   My server crashed today morning due to PSU failure, and now it is
> checking (in foreground!) 8Tb UFS2+SU volume for 6200 seconds, and it
> is only "Phase 1b" (!!!). I don't want even think about background
> check of this FS.
> 
>   Is SUJ stable enough to migrate to it? It was marked as stable some
> time ago, and was included into 9-RELEASE, but later I seen some
> messages on fs@ list, that it still has some problems, and even some
> references to McKusick's message about this instability (but I've failed
> to find message itself).

Most of the issues with SUJ are related to their interaction with
snapshots. At the moment 9-head (head of the 9 branch) has had
the taking of snapshots disabled on filesystems running with SUJ
(but not with SU). There have been some important bug fixes to SUJ
since 9-release. So if you wish to use SUJ, I recommend using
9-head rather than 9-release.

>   BTW, this check reveals many softupdate inconsistences (mostly DUPs),
> and most of them are in files, which was not written for sure in time
> of crash (old archives, which could be only read!).
> 
> --
> // Black Lion AKA Lev Serebryakov <lev at FreeBSD.org>

It sounds like you have a disk sector containing a block of inodes
trashed or gone bad.

	Kirk McKusick


More information about the freebsd-fs mailing list