Re: 2.4DEVEL -> 2.4STABLE?

On Fri, Oct 20, 2000, Robert Collins wrote:
> I mailed a patch to squid-dev early this week
>
> The current code detects
> *missing cache files
> *files with inconsistent lengths - disk size <> swap.state expected size
>
> and then asserts if any errors were found

uhm .. to make this clearer - the current code doesn't delete the bad
swapfiles (or perform any other corrective action) so it sits in a
tight loop.

> The patch changes the code to
> *remove missing cache files from the swap state
> *remove and unlink from disk files with inconsistent lengths
>
> and then debug the number of errors found at the end of validation.

yup.

> It hasn't been tested in large scale production caches however. I have
> checked it does the job in small scale stuff.

Ok, the original bug was found by Kinkie. Were you able to test the patch
out on your box at all kinkie? It'd be nice to know if it worked or not
on a usefully sized box. I've tried it here but since my cache isn't
corrupted .. :-)

Thing is, squid -S wasn't documented in the usage() when I did my initial
modio work which is why I added it. Hrm, I wonder who actually DID add it
and whether I just screwed up when I broke th code out into per-storeFS..

Adrian

-- 
Adrian Chadd			The Law of Software Development and
<adrian@creative.net.au>	Envelopment at MIT:
    "Every program in development at MIT expands until it can read mail."