[2027] in Release_7.7_team
attach and hard vs. soft NFS mounts
daemon@ATHENA.MIT.EDU (Abby Fox)
Mon Jan 10 15:51:30 2000
Message-Id: <200001102051.PAA26493@whirlwind.mit.edu>
To: release-team@MIT.EDU
Date: Mon, 10 Jan 2000 15:51:09 -0500
From: Abby Fox <ajfox@MIT.EDU>
If we're having a meeting this week, can we talk about how attach
behaves wrt hard/soft NFS mounts? (This is something that came up when
Bob and Miki were helping to troubleshoot a file corruption problem
which we hope the IRIX patch in 8.3.24 will fix, but I'd like to be
clear on things in case it comes up again.)
Specifically, Solaris and other docs on NFS say not to use a soft
mount for read-write because:
Applications frequently do not check return values from soft mounted
file systems, which can make the application fail or can lead to
corrupted files. Even if the application does check, routing
problems and other conditions can still confuse the application or
lead to file corruption if the soft option is used.
So I assume there's a good reason why we have soft as default, but my
other question is whether "attach -o hard" actually overrides it,
since the mnttab lists both:
sgi% attach -o hard imagery4
attach: IMAGERY.MIT.EDU:/u4 attached to /mit/imagery4 for filesystem imagery4
sgi% mount | grep imag
IMAGERY.MIT.EDU:/u4 on /mit/imagery4 type nfs (vers=3,size=1024,wsize=1024,soft
,rw,hard,nosuid,nodev,dev=100001)
sun% attach -o hard imagery4
attach: IMAGERY.MIT.EDU:/u4 attached to /mit/imagery4 for filesystem imagery4
sun% mount | grep imag
/mit/imagery4 on IMAGERY.MIT.EDU:/u4 wsize=1024/soft/read/write/hard/nosuid/remote on Mon Jan 10 15:39:50 2000