[OpenAFS] humungous disk caches

[Neulinger, Nathan]

It doesn't work. You just haven't caused the kernel to flush an inode
yet. The moment you do - bang!

Trust us, it does not work. Use ext3, it works fine. All of my machines
are reiserfs based, except for the afs cache.=20

-- Nathan

------------------------------------------------------------
Nathan Neulinger                       EMail:  nneul@umr.edu
University of Missouri - Rolla         Phone: (573) 341-4841
Computing Services                       Fax: (573) 341-4216


> -----Original Message-----
> From: Bryan Bayerdorffer [mailto:bryan.bayerdorffer@spd.analog.com]=20
> Sent: Wednesday, March 12, 2003 12:48 PM
> To: openafs-info@openafs.org
> Subject: Re: [OpenAFS] humungous disk caches
>=20
>=20
> So far it seems to work though.  Curious.  I chose that=20
> mainly for its good=20
> performance with large numbers of files.
>=20
> How about ext3?  Does the journaling get in the way at all? =20
> Is ext2 the only=20
> safe option?
>=20
> Derrick J Brashear wrote:
> > On Wed, 12 Mar 2003, Bryan Bayerdorffer wrote:
> >=20
> >=20
> >>Setting aside the question of whether a huge cache is a=20
> good idea, what's the
> >>largest client disk cache that can be configured?  I can=20
> get 16GB so far with
> >>-chunksize 18 and -files 65536, but a chunksize of 19=20
> causes accesses to /afs
> >>to hang, as does doubling the number of files (cacheinfo is=20
> always set
> >>correspondingly).
> >>
> >>What causes the hangs?  I'm using -dcache 3600 right now. =20
> Does this value
> >>correlate to the cache size in any way other than performance?
> >>
> >>The clients are openMosix cluster nodes, and the cache is=20
> on reiserfs with -o
> >>nolog,notail.
> >=20
> >=20
> > we don't support reiserfs for cache, period, since it=20
> doesn't bother to
> > support unique inode numbers.
> >=20
> >=20
> >=20
> > _______________________________________________
> > OpenAFS-info mailing list
> > OpenAFS-info@openafs.org
> > https://lists.openafs.org/mailman/listinfo/openafs-info
> >=20
>=20
> --=20
>   .. ..-. ..- -.-. .- -. .-. . .- -.. - .... .. ... --. . -=20
> .- .-.. .. ..-. .
> Bryan Bayerdorffer        bryan@meatspace.net          =20
> bryan@spd.analog.com
>                       (Wit's End Computation Center)      =20
> (Analog Devices)
>=20
> "Man's chief occupation is extermination of other animals and his own
> species, which, however, multiplies with such insistent rapidity as to
> infest the whole habitable earth and Canada."
>=20
> _______________________________________________
> OpenAFS-info mailing list
> OpenAFS-info@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-info
>=20