Date: Thu, 5 Mar 1998 23:02:01 -0800 (PST) From: Tom Bartol <bartol@salk.edu> To: dyson@FreeBSD.ORG Cc: karl@mcs.net, shimon@simon-shapiro.org, current@FreeBSD.ORG, jak@cetlink.net, karpen@ocean.campus.luth.se Subject: -current NFS discussion Message-ID: <Pine.BSF.3.96.980305222353.850D-100000@dale.salk.edu> In-Reply-To: <19980305215754.02721@mcs.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello John et al., Just to add my own observations of -current NFS... We run a mixture of unix platforms at our site (SunOS 4.1.3U1, IRIX 5.X and 6.X, Digital UNIX 4.0, Linux 2.X, FreeBSD 2.2.5 and -current) and our NFS server is an Auspex NS7000. I support -current on my own box here so that I can keep up-to-date with things and advise the rest of the lab on progress. My machine mounts the Auspex in NFSv3 mode at 100Mb/s across an ethernet switch. I have an Intel EtherExpress Pro 100b interface card. Stability of NFS has been shakey over the past several weeks so I keep having to fall back to the Feb 14 kernel I built. Although today's kernel (cvsupped the morning of March 5 PST) seems to be working fine so far (I believe the recent change to vfs_bio.c v1.153 on March 4 may be responsible for this for me). However, I am still experiencing (i.e. I've reported this before) an NFS-related problem with today's -current whereby if I use the the compat-linux gcc to compile source code located on an NFS mounted filesystem the resulting executable is defective and will not run. The same code compiles perfectly when located on a local filesystem. For some reason, trivially simple code (like hello_world) compiles fine even if across NFS but larger code fails. I have not been able to narrow down the point in complexity or file-size at which the problem creeps in but I'd say the evidence points at a problem in NFS still lurking in there. So, you may ask, why do I care about compiling Linux binaries on a FreeBSD box? Well, believe it or not, it's very useful to us because we make great use of Matlab (for which there is a Linux version but not a FreeBSD version) and we need to link in our own object code to operate within Matlab efficiently. Pretty twisted, I know, but it does work rather well (except for the NFS problem, that is). If I can be of help to supply more info on this problem please let me know. Many, many thanks, Tom On Thu, 5 Mar 1998, Karl Denninger wrote: > On Thu, Mar 05, 1998 at 10:46:30PM -0500, John S. Dyson wrote: > > Karl Denninger said: > > > On Thu, Mar 05, 1998 at 10:34:43PM -0500, John S. Dyson wrote: > > > > Simon Shapiro said: > > > > > > > > > > If NFS is important, then let's stabilize what we do, plan the project > > > > > carefully, divide the work up and do it. This will mean putting all other > > > > > development into priority-1-fixes-only until NFS is done. It is more than > > > > > one person's weekend job. > > > > > > > > > > I'll take a piece of such effort. > > > > > > > > > I think that you'll be pretty happy with NFS by Fri night :-). I am working > > > > no-holds-barred on it right NOW. > > > > > > > > -- > > > > John | Never try to teach a pig to sing, > > > > dyson@freebsd.org | it just makes you look stupid, > > > > jdyson@nc.com | and it irritates the pig. > > > > > > Client and server? > > > > > Mostly client. Server should be mostly a matter of fixing whatever recent > > breakage has happened. > > > > -- > > John | Never try to teach a pig to sing, > > dyson@freebsd.org | it just makes you look stupid, > > jdyson@nc.com | and it irritates the pig. > > I don't know if there has been any recent breakage.... does anyone else have > any serious (ie: heavy-duty) experience in this area that can speak to > whether the -CURRENT tree, as it exists now, is ok in the NFS server area? > > -- > -- > Karl Denninger (karl@MCS.Net)| MCSNet - Serving Chicagoland and Wisconsin > http://www.mcs.net/ | T1's from $600 monthly to FULL DS-3 Service > | NEW! K56Flex support on ALL modems > Voice: [+1 312 803-MCS1 x219]| EXCLUSIVE NEW FEATURE ON ALL PERSONAL ACCOUNTS > Fax: [+1 312 803-4929] | *SPAMBLOCK* Technology now included at no cost > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-current" in the body of the message > 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?Pine.BSF.3.96.980305222353.850D-100000>