Re: [SQU] [From CVS] cache no longer working ...

I've switched this to squid-dev (if it ain't obvious :0)

I confess to simply taking the existing logic and making it 'repair rather
than die'.

The background check (which I should get onto very soon) will be working
from the in memory swap.state so the logic should be correct.

I thought Duane's key gripe was - it slows down cache response time and time
to first hit. All to check stuff that is checked on each hit anyway...

The question I have is - at what debug level should these messages be spat
out in the background task version?

Rob

----- Original Message -----
From: "Adrian Chadd" <adrian@squid-cache.org>
To: "Robert Collins" <robert.collins@itdomain.com.au>
Cc: <squid-users@ircache.net>
Sent: Wednesday, November 22, 2000 8:52 AM
Subject: Re: [SQU] [From CVS] cache no longer working ...

> On Wed, Nov 22, 2000, Robert Collins wrote:
> > I take it you'd like the background-check version to have a higher debug
> > level? :-]
> >
>
> Well, why should it complain about a size mismatch or missing file
> when rebuilding from a log file? if the cache shut down in a dirty
> fashion, the log file will be a log of operations, with adds and
> deletes, so most of the file *will* be invalid. :-)
>
> I think this is what Duane was getting at on squid-dev.
>
> The logic is fine for a clean log rebuild or a rebuild-from-storedir,
> but not for a dirty log rebuild.
>
>
> Adrian
>
>
Received on Tue Nov 21 2000 - 14:58:25 MST