Date: Thu, 27 Apr 2000 18:52:47 +0200 From: Brad Knowles <blk@skynet.be> To: Matthew Dillon <dillon@apollo.backplane.com> Cc: "John W. DeBoskey" <jwd@unx.sas.com>, freebsd-current@FreeBSD.ORG Subject: Re: Support for large mfs Message-ID: <v04220815b52e1ede90ac@[195.238.1.121]> In-Reply-To: <200004271634.JAA05279@apollo.backplane.com> References: <200004270554.BAA34693@bb01f39.unx.sas.com> <200004270605.XAA00807@apollo.backplane.com> <v04220800b52da49fcdee@[195.238.23.59]> <200004271634.JAA05279@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
At 9:34 AM -0700 2000/4/27, Matthew Dillon wrote: > I can't imagine why MFS would perform better... it shouldn't, every > block is stored in system memory *TWICE* (once in the VM cache, and > once in the mfs process's address space). If you have enough system > memory to create a large MFS filesystem and it performs well, then > the system should perform even better if you remove the MFS filesystem > and just use a normal filesystem. When I tried using a regular file on the filesystem, my system time went way up, my iowait went way up, and my performance dropped through the floor. However, this was on 3.2-RELEASE and was several months ago -- I hope that 4.0-STABLE would be a bit better about that. ;-) > I would consider trying a normal filesystem with an async or a >softupdates > mount. Or a normal filesystem with softupdates enabled. It may also > help to turn off write-behind (sysctl -w vfs.write_behind=0), though if > you are running the latest 4.x stable the write heuristic is now in and > should do a good job on its own. I believe I had tried this with softupdates at the time, but it's been long enough since I tried this that I can't be sure. I do recall seeing my performance go down quickly enough and seeing the disk I/O go through the roof fast enough that I decided I would never, ever, ever try that again. Of course, now I am revisiting that decision. ;-) I'm seriously contemplating getting a Dell PowerEdge 2450 with five internal 10kRPM/18GB disks, 2GB of RAM, two of the fastest processors they've got, perhaps a pair of Intel EtherExpress Pro 100+ NICs, and giving this another go with FreeBSD 4.0-STABLE. Do you have any thoughts on this subject? Is 3.2-RELEASE old and non-optimized enough that it really could stand replacing? Given some of the problems we've been seeing lately (which I'm pretty sure are a result of running out of virtual memory and the machine spontaneously rebooting), I think I might be able to convince my management to spring for a third 2450, but with a slightly different configuration than what I'll be using for the news reader servers. It would help me formulate useful arguments for this proposal if I had input from more knowledgeable people than I. -- These are my opinions -- not to be taken as official Skynet policy ====================================================================== Brad Knowles, <blk@skynet.be> || Belgacom Skynet SA/NV Systems Architect, Mail/News/FTP/Proxy Admin || Rue Colonel Bourg, 124 Phone/Fax: +32-2-706.13.11/12.49 || B-1140 Brussels http://www.skynet.be || Belgium To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?v04220815b52e1ede90ac>