Re: Re-checking disk cache

On Tue, Oct 10, 2000, Chemolli Francesco (USI) wrote:
> Today I had a rather spectacular system crash.
> I quickly recovered (God Bless ReiserFS), but then
> after a while my squid CPU usage shot up to 100%,
> and I got messages of "WARNING: swapin MD5 mismatch".
> But the cache sort of worked, so I left it run.
>
> Now at business-day-end, I was able to shut down
> squid and restart it to double-check the disk caches.
> Lots of objects are broken, it's taking forever.
>
> This got me thinking. How hard would it be to
> implement an option to force a disk cache re-check while
> the cache is live? (read: could pretty please somebody
> do that?)

No idea. Thing is, I know that *I* am in code tidyup and general bugfix mode,
so I don't think I will be able to add any new features to squid just yet.
Duane, can you email it to the "squid wishlist" please ?

> And then some more: I'm using Linux, and I'm an happy
> diskd user. I've heard voices of aufs being potentially faster
> on Linux, but I can't use it, the SIGUSR* would be
> lost and I _need_ logfile rotation.
> How hard would it be to implement some other scriptable
> control interface? For instance, a named pipe would be
> quite OK with squid's overall model, and it would cost only
> one FD.

Uhm, you mean a control channel to do logfile rotation? Hrm. I think
that might be a candidate for a cachemgr + acl option. Look t adding
a cachemgr option to do it, it can't be that difficult (look at the
code that toggles the offline mode to see how to implement the cachemgr
interface) and send patches to -HEAD this way.

Adrian

-- 
Adrian Chadd			"If a butterfly flaps its wings in China,
<adrian@creative.net.au>	    will a woman get naked in Amsterdam?"
				      -- Ashley Penney on Chaos Theory