Subject: Re: ffs_valloc: dup alloc
To: None <port-i386@netbsd.org>
From: Stephen Borrill <netbsd@precedence.co.uk>
List: port-i386
Date: 02/26/2002 11:58:22
> On Mon, Feb 25, 2002 at 05:22:16PM +0000, Christos Zoulas wrote: In article
> > <<EMAIL: PROTECTED>>, Stephen Borrill <<EMAIL: PROTECTED>> wrote: I've a
> > > customer with a 1.4.3 machine (upgrading to 1.5 is not an option at the
> > > moment). They are running crashed quite often with a squid as a web
> > > proxy. The machine panics quite often with ffs_valloc: dup alloc (it's
> > > squid it panics in). Thinking this was a bug in the FFS code in 1.4.3
> > > that fsck could not fix, I newfs'd the partition (it's basic ffs) and
> > > restored from a backup. It's still happening though. What's the best
> > > way to work around this problem? iclr and then fsck?
> > >
> >
> > I've seen that too. There was a problem in 1.4.x with triple-indirect
> > blocks and truncation that squid used to trigger and charles fixed. I
> > still see
> 
> A simple way to avoid this problem is to increase your filesystem
> blocksize; then you won't *have* to use triply-indirect blocks (thank
> goodness).

That sounds like a good plan. The current filesystem block size is 8192 and
the partition in question has a size of 37846368 512-byte sectors. What's the
rule of thumb to calculate the minimum block size that will stop
triple-indirects given such parameters?

-- 
Dr. Stephen Borrill