Re: Has any one got AFS server/client sucessfuly working on the powerpc line?
- To: debian-powerpc <email@example.com>
- Subject: Re: Has any one got AFS server/client sucessfuly working on the powerpc line?
- From: Michael Furr <firstname.lastname@example.org>
- Date: 26 Sep 2002 17:58:32 -0400
- Message-id: <1033077513.25666.109.camel@apocalypse>
- In-reply-to: <1033072964.3286.4.camel@asterix>
- References: <1032980094.15533.4.camel@asterix> <1033007576.11778.10.camel@orthrus> <1033072964.3286.4.camel@asterix>
On Thu, 2002-09-26 at 16:42, Adam wrote:
> Thank-you. Yes, I have been looking at openAFS and reading all their
> webpages and documentations. Is it best to run openAFS with xfs? I am
> using ext3 right now and look to be switching to xfs in the future. I
> have not got xfs compiled in my kernel just yet (2.4.18).
After going back and looking again, we are actually using xfs(kern
2.4.17) for all of our vice partitions on the linux side, so it seems
whatever problems I recall having(this was a while ago) have been
> Also, does it mater for efenciently the processor I use on the openAFS
> servers or is the memory and storage the most important functions of the
> hardware I choose to use to serve the openAFS cell/server?
Yes, the processor speed is NOT the bottleneck. Network bandwidth and
memory are the things to load up on. Our busiest file server is a Sun
Netra T1 that has a consistant load of about .5. Also for where its
applicable, read-only replication really helps out performance for
heavily used volumes. For instance, we link all of our workstation's
/usr/local's into /afs and replicate this, which works really well.
And due to its distributed nature, lots of smaller servers give you much
more flexability than a single large one...