home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.ultrix
- Path: sparky!uunet!cs.utexas.edu!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!ira.uka.de!fauern!LRZnews!petav
- From: petav@Physik.TU-Muenchen.DE (Peter Averkamp)
- Subject: Re: Slow writes to Ultrix 4.3 NFS server
- Message-ID: <petav.722282161@e20.physik.TU-Muenchen.DE>
- Keywords: NFS server performance
- Sender: news@news.lrz-muenchen.de (Mr. News)
- Organization: Leibniz-Rechenzentrum, Muenchen (Germany)
- References: <leb.721702653@Hypatia> <Bxos0B.GqM@news.iastate.edu>
- Date: Fri, 20 Nov 1992 17:56:01 GMT
- Lines: 78
-
- john@iastate.edu (John Hascall) writes:
-
- >leb@Hypatia.gsfc.nasa.gov (Lee E. Brotzman) writes:
- >}I manage an SGI 4D/340 running IRIX 4.0.1 and a DECstation 5000/125 running
- >}Ultrix 4.3. The user login directories of both machines are NFS mounted to
- >}each other. I have found that when the DEC box writes files to the SGI,
- >}performance is good, however when the SGI tries to write to the DEC, throughput
- >}is just awful.
-
- >Is this in the FAQ? If not it should be...
-
- >Anyway, several vendors' NFS servers mis-implement the standard (on purpose).
- >The NFS standard requires that writes be done "synchronously". Some vendors
- >allow you to set an option which makes NFS writes be done "asynchronously"
- >which can gain you a massive performance increase -- but if your server has
- >a catastrophic crash (and can't flush the disk buffers) you can have data
- >integrity problems. *** SGI ships their systems with ASYNCH as the DEFAULT ***
- >If you run asynch-writes (and we have modified our DEC servers to do this) and
- >you don't have them on a UPS you are running a massive risk.
-
- >The "safe" option to make your Ultrix boxes faster is to buy PrestoServe
- >for your NFS servers.
-
- >John
- >(if you have an Ultrix *source* license, I can provide the 1-line patch)
- >--
- >John Hascall ``An ill-chosen word is the fool's messenger.''
- >Systems Software Engineer
- >Project Vincent
- >Iowa State University Computation Center + Ames, IA 50011 + 515/294-9551
-
- There has been a long discussion about all this over and over some time
- ago. Being tired of people talking about things many of them never have
- tried out, i have put together an object-code patch for the versions
- 4.2 and 4.2a, so more people could try and take the risk and make their
- own experiences. I have mailed it to about 20 sites with a pledge to
- inform me if >>anything<< >>ever<< goes wrong with a patched server. Up
- to now, all i got were thanks from happy users. I know that 20 is very
- bad statistics, so I'd like to see if there are even more dare-devils
- out there taking this data-loss risk...
- We don't have 4.3 yet, but you might try to use mine or figure out your
- own patch. By the time we get 4.3, i'll update it; maybe some other
- kind soul will do this before. I have binary patches for VAX/Ultrix(4.1)
- also.
-
- Now here's what you might want to do:
-
- cd /sys/MIPS/BINARY
- cp nfs_server.o nfs_server.o.orig
- uuencode nfs_server.o.orig nfs_server.o|sed -e 's/P$ \!B0\!/P$ \!B0 /'|uudecode
-
- the patched nfs_server.o's sums:
- sum for 4.2 should be: 62555 104
- sum for 4.2a should be: 15177 104
- (The 'sum' of the uuencode-line is:18397 1, obtained by echo -n "stuff"|sum)
- And finally, please mail:petav@e20.physik.tu-muenchen.de simply
- saying that you've done it and if or what went wrong. I am really
- interested in getting better statistics to hear about anything that
- ever went wrong, caused by the patch.
-
- Please,Please be aware that using this will kick you completely off
- limits; do not complain about anything to your dec service person if
- you use a patched system. Do not blame or flame me! This is a
- completely experimental modification that's only there to amuse those
- who play with it. No Warranties, not even those for failure.
-
- Many thanks to John Hascall who kicked off this whole issue and was so
- kind to verify the patches against a source-patched object.
-
-
- Peter Averkamp
-
-
- --
- Peter Averkamp, | email:
- Physics Department E20 | petav@radon.e20.physik.tu-muenchen.de
- Techn. Univ. of Munich | Phone: ++49 (89) 3209-2408
- D-8046 Garching, Germany | Fax: ++49 (89) 3209-2338
-