Re: Hmmmm.

It'd probably be possible to use the inode hack for news servers on linux
to improve squid as well. The hack where it provides a mechanism to get
an inode directory without a filename. Then squid needs to keep track of
the inode in its (presumably) more efficient method than the fs.

See <http://www.ecsnet.com/html/20-p0083.html>.

Dean

On Wed, 19 Mar 1997, Michael O'Reilly wrote:

> In message <Pine.BSF.3.95.970319093026.8888K-100000@skylark.hilink.com.au>, "Da
> niel O'Callaghan" writes:
> > If unlinking expired objects is time consuming because of the constant
> > inode metadata updates, it might be an idea to suggest to people with busy
> > caches to 'mount -u -o async /cachedir' on those OSs that support it.
> > Linux does not need this, apparently, because ext2fs defaults to async
> > mode. Async mode causes the inode metadata to be written to disk at the
> > next sync(), rather than immediately.
>
> Ummm. those timings were taking on a linux system, that also had the
> cache partitions mounted with 'no_atime'. (i.e. don't update the
> access time in the inodes) which further reduces the amount of data
> written.
>
> I'm not 100% sure why it's so slow, but I suspect is just that the
> number of directories is high enough, and the access random enough
> that the inode caching fails, and it ends up frequently waiting for
> inodes to be loaded.
>
> Michael.
>
>
Received on Tue Jul 29 2003 - 13:15:40 MDT