Date: Mon, 13 Nov 1995 08:26:53 -0800 From: David Greenman <davidg@Root.COM> To: Simon Marlow <simonm@dcs.gla.ac.uk> Cc: uhclem (Frank Durda IV) <@fw.ast.com:uhclem@nemesis>, current@freebsd.org Subject: Re: Disk I/O that binds Message-ID: <199511131626.IAA04123@corbin.Root.COM> In-Reply-To: Your message of "Mon, 13 Nov 95 13:42:45 GMT." <199511131422.GAA06017@who.cdrom.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> >> My system where I have seen something that somewhat resembles his >> description is a mixed IDE SCSI system, and when things bog, the hog >> process and the binding process are both trying to get to the same >> SCSI drive. The IDE is for swap and root and is idle during these events. >> >> And in answer to someone elses' question, there was no tape I/O or >> CD-ROM I/O active during these events on either system. Only SCSI disk. > >Just a guess, but could this be caused by the disk request sorting >done by FreeBSD? New requests get entered into the queue possibly >before old ones, and it could be a long time before a request finally >gets serviced. Yes, and this does contribute to the problem. This was the topic of a recent discussion between myself and John Dyson. Our disksort could use a little work. One thing I'd like to do is make it possible for the driver to request that no sorting be done. This would allow controllers that can do tagged queuing to pass the I/O off to the drive unordered - allowing the drive to use superior sorting algorithms. -DG
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199511131626.IAA04123>