Date: Wed, 8 Nov 2000 19:27:20 -0800 (PST) From: Brian Behlendorf <brian@collab.net> To: stable@freebsd.org Subject: Re: disk I/O Message-ID: <Pine.BSF.4.21.0011081851080.1844-100000@yez.hyperreal.org> In-Reply-To: <Pine.BSF.4.05.10011081745020.11646-100000@shell.uniserve.ca>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 8 Nov 2000, Alfred Perlstein wrote: > You have an IO problem that leads to a cascade failure, basically > everything starts piling up on the disks, Yep, that's basically what I'm seeing. > I'd get a hardware raid system, they're a lifesaver for IO and > reliability. I.e., the conclusion is that my load is greater than what the hardware can support; that's confirmed by: On Wed, 8 Nov 2000, Tom wrote: > A 100tps average is about 10ms per transfer. A good hard drive takes > about 8ms average per seek. Assuming every transfer requires a seek, you > are not doing too badly. However, tagged commands allow multiple pending > IOs per disk. If tags are disabled, or not functioning you will not get > this benefit. See camcontrol. Tags are enabled according to camcontrol; and yep, these should be ~8ms disks, so anything over 125 I'm getting the benefits of tagging. Maybe what I'm looking for is a way to increase the amount of RAM cache I can make available for disk accesses; what's the standard place to look at for monitoring & tuning that? Or is this something that it's better not to touch, just let the OS figure out? I ask because I don't appear to be using all the RAM I've got available (based on the fact that I hardly ever swap). RAID is an option, of course. I looked at Mike Tancsa's #'s for the 3Ware driver and they were *quite* nice. I was just hoping I'd get more out of this current hardware. No luck! For the sake of completeness: On Wed, 8 Nov 2000, Simon wrote: > Is that machine using any swap when this happens? If so, that's most > likely why. It tries to spawn processes, but can't do it fast enough. > It happens on my box now and then. Running 4.1-R Barely touches swap (according to swapinfo, as well as pi/po in iostat) On Wed, 8 Nov 2000, Mike Tancsa wrote: > Also, mounting your disks with noatime (mount -u -o noatime /blah) > requires no downtime and immediately improvement for a lot of things. > Basically, it elimiates an IO to update the atime on files and > directories. > Also, softupdates will be very helpful, but will probably require a > reboot. Using both softupdates & noatime. =) Thanks for the feedback, everyone! Brian To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0011081851080.1844-100000>