On Thu, Sep 27, 2001, Mike Diggins wrote:
>
> I think that someone just asked this question but I've been meaning to ask
> it myself. I'm running Squid 2.4S1 on Solaris 8 and using Diskd. My cache
> has been at about 90% for sometime now. I'm curious why my 'calls to
> unlinkd' always says zero? Something must have been deleted from the cache
> by now. Is there some other process doing this or is something wrong?
The unlinks are handled through diskd, rather than an unlinkd process.
If you were using the normal 'ufs' storage type, unlinks would be
processed by the unlinkd, since the ufs file operations are all sync
(vs async with diskd/aufs).
I also believe that if the cache _requires_ a whole heap of objects
to be deleted right now to make way for new ones (unlikely - this'll
generally happen if you decide to shrink your storage, or your
squid is really really really busy) the diskd code will actually
make calls to unlinkd to delete the files as a last result.
HTH,
Adrian
Received on Thu Sep 27 2001 - 12:11:21 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:02:30 MST