Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 22 May 2005 10:40:50 -0300 (ADT)
From:      "Marc G. Fournier" <scrappy@hub.org>
To:        freebsd-questions@freebsd.org
Cc:        freebsd-stable@freebsd.org
Subject:   [4.11-STABLE] sporatic directory corruption with unionfs ...
Message-ID:  <20050522101434.W61528@ganymede.hub.org>

next in thread | raw e-mail | index | archive | help

I have 6 servers in place right now, with the 6th being a brand new Intel 
1U box, using a 3Ware 9500 controller to do RAID5 for 3xSATA drives ... 
yesterday, for the first time, I've started to get a sort of file system 
corruption that I've never seen before ... basically, it looks like the 
directory entries are getting corrupted, but the files themselves are fine 
...

Basically, I have a jail, with /usr in the jail being a unionfs from a 
template ... so:

mount_union -b /template/usr /jail/usr

in the template, there is a directory /usr/local/cyrus/bin, for instance 
... the binaries under /usr/local/cyrus/bin are literally disappearing, as 
if someone were entring that directory and doing a 'rm' of those files ...

I have a serial console attached to this server, and there are no errors 
being reported by the operating system ... tw_cli is showing that drives 
and unit are both functioning properly ... so I can find no reasons for 
the apparent "corruption" ...

I just brought the server down to single user mode, and umounted 
everything so that I could do an fsck on the file system itself, and *its* 
checking as being clean ... its still running though, so there may be 
something at the end ...

When I first noticed this, yesterday, the server had been running 11 days, 
and nothing in /var/log/messages to indicate a problem ... today, after 
<24hrs, its done it again, and again, no errors being generated anywhere 
to indicate a problem ...

All the other 5 servers are running SCSI ... but unless there is a bug in 
the 9500 driver, I can't see it being hardware related ...

I've kinda always expected something like this might happen with unionfs 
as a result of the server crashing, but not when its running fine ...

I don't know what else to add to this, unfortunately, since neither the 
hardware, or the operating system, seem to want to give anything up :(

The OS is from April 11th, so is relatively recent ...

Any suggestions/ideas on what else to look at would be much appreciated 
...

Thank you ...

----
Marc G. Fournier           Hub.Org Networking Services (http://www.hub.org)
Email: scrappy@hub.org           Yahoo!: yscrappy              ICQ: 7615664



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050522101434.W61528>