Subject: Re: is something wrong with either ccd(4) or isp4(4), maybe only on alpha?
To: NetBSD Kernel Technical Discussion List <tech-kern@NetBSD.org>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: tech-kern
Date: 10/25/2004 18:35:45
> However there's something weird going on with read rates on the CCD
> driver on my alpha.  [...]

In a past life I was a sysadmin at a university site.  We had an alpha
which was intended to be a backup server (its raison d'être was to run
amanda, basically).  But performance was pitiful - and I mean _really_
pitiful, the sort that leads to things like "yes I said kilo not mega".

It could send and receive network data at full wire speed.  It could
read and write the disk zippily.  But as soon as I tried to do both,
performance plummeted.

Now, in that particular case it turned out to be brokenness in the PCI
management logic (interrupts or bus mastery or something, I don't
remember); any one thing worked fine, but try to overlap and it got
horribly confused.  This doesn't sound like your problem - but it
occurs to me that what one company can botch, another company can botch
differently.  Perhaps your machine's hardware does not get along well
with attempts to load multiple disks heavily simultaneously?

> Despite this I'm still more inclined to believe the problem is with
> interleaving in the ccd(4) driver on the alpha than it being a
> problem in the isp(4) driver.  I say this because a single drive
> doesn't seem to exhibit the same problem, nor does a plain
> concatenation of the four drives with ileave=0.

This would also be consistent with my theory above.  Maybe try
simultaneous dd reads/writes to/from /dev/{null,zero} to multiple
drives, without ccd involvement?  That should tell whether it's simply
multi-drive access that's killing it (vs ccd trouble).

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B