Date: Sat, 25 Jan 2003 01:29:22 -0500 From: Bill Moran <wmoran@potentialtech.com> To: Kevin Stevens <Kevin_Stevens@pursued-with.net> Cc: freebsd-questions@freebsd.org Subject: Re: Filesystem tuning parameters Message-ID: <3E322EC2.7000506@potentialtech.com> References: <2F48C1C3-3022-11D7-8DC1-003065715DA8@pursued-with.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Kevin Stevens wrote: > > On Friday, Jan 24, 2003, at 16:40 US/Pacific, Bill Moran wrote: > >> See >> /usr/share/doc/papers/diskperf.ascii.gz >> on your system. This is the authoritative resource as to why those >> settings >> are they way they are. > > ?? Sure that's the correct doc? It involves throughput tests of > different disk systems on VAXen, but doesn't really discuss any of these > parameter changes. They do go into rotational delay a bit. Hmmm ... perhaps I'm wrong. I thought that was it, but I remember more information about the testing that led to decisions about the way the filesystem works. >> So ... it's like this: >> 1) If you really want to fill your drive up past 90%, understand that UFS >> simply isn't designed to do that efficiently. > > Ok... and what you're confirming is that this is a percentage > requirement, so it doesn't vary significantly between 120MB and 120GB > filesystems? Yes. While I don't understand the deep magic of it, the fact is the amount of free space needed to ensure efficient block allocation is a percentage of total filesystem space. >> 4) If you bought a 120G drive because you have 119.5G of data to store, I >> think you made a mistake and should either return it for a bigger >> drive >> or accept the performance hit. > > My confusion came from various bits of documentation that suggest the > primary purpose of minfree is to provide notification and buffer > time/space for sysadmins to deal with filesystems nearing capacity. Well, fact is you _can_ fill a disk past the 92% mark. But as a sysadmin, you'll definately want to be alerted to this because the write algorithm changes from time to space and performance drops dramatically. > In > my scenario, 12GB would be total overkill to commit for that purpose, > regardless of how much data I needed to store. Understanding that it is > required for filesystem overhead makes the resource usage justifiable. > Thanks! I see where you're coming from. Glad I could help clear it up. -- Bill Moran Potential Technologies http://www.potentialtech.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3E322EC2.7000506>