5-STABLE softupdates issue?

[Scott Long]

Michael Nottebrock wrote:
> Matthias Andree wrote:
> 
> 
>> I posted about softupdate problems on a SCSI system with DISABLED WRITE
>> CACHE, on a somewhat flakey Micropolis drive that froze and caused
>> massive ffs+softupdates corruption in February 2004 (on FreeBSD 4
>> though), see <URL:http://docs.freebsd.org/cgi/mid.cgi?m38yj15m59.fsf>
>> for the archived post, including logs.
> 
> 
> FWIW, I'm used to "UNEXPECTED SOFTUPDATE INCONSISTENCY" erros as well, 
> been getting them every now and then after the occasional system lockup 
> or kernel panic as long as I've been running freebsd and I'm pretty sure 
> with each release since 4.0-R, too. I didn't know those are _literally_ 
> unexpected. If so, then softupdates is a probably lot more flaky than it 
> ought to be.
> 

It's common (and even expected) for fsck to clean up normal things like 
unlinked files after a hard shutdown when softupdates is enabled. 
However, if fsck is bailing out and telling you that there is an
unexpected inconsistency that needs to be fixed manually, that is not
normal.  Unfortunately, most of the evidence is probably long gone by
then, so tracking it down is pretty tricky.  I've had talks with some
filesystem guys at a well-known Unix company who have talked of evil
things like extracting dirty buffers from a crashdump file; that may
be the level of effort needed to track down these kinds of problems.

Scott