Bug 7947

Summary:	sed breaks on and NFS partition
Product:	[Retired] Red Hat Linux	Reporter:	Ken Filipps <kfilipps>
Component:	knfsd	Assignee:	Jeff Johnson <jbj>
Status:	CLOSED WORKSFORME	QA Contact:
Severity:	medium	Docs Contact:
Priority:	low
Version:	6.1
Target Milestone:	---
Target Release:	---
Hardware:	sparc
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2000-02-17 15:45:01 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Ken Filipps 1999-12-22 14:25:56 UTC

This is a strange one...

I discovered while compiling a few packages for a new RedHat 6.1 (Sparc)
system I was setting up that running a 'configure' script on a piece of
software that the sed substitutions would fail when it was run on a NFS
mounted partition.  There was no problem on a local drive.  I have found
this with all packages I have tried.

However, I can read and write with no problems to the NFS mounted
partition.  There seems to be no issue with the partition.  I mount my
partitions from a Solaris 2.6 (Sparc) box.

Here is the tail end of a failed attempt (using screen-3.9.5 as an
example):

-----
updating cache ./config.cache
creating ./config.status
creating Makefile
sed: file conftest.s1 line 1: Unknown command: ``^''
creating doc/Makefile
sed: file conftest.s1 line 1: Unknown command: ``^''
creating config.h
sed: file conftest.frag line 1: Unknown command: ``%''
-----

I am stumped as to why these things should even be related...

My system is RedHat 6.1 with all relevent updates applied and my kernel is
a 2.2.12-42 stock kernel from the original install.

Comment 1 Jeff Johnson 1999-12-22 16:41:59 UTC

If your promlem with sed disappears on local storage, but appears on NFS
mounts, then the problem is not with sed at all. NFS can and does inject blocks
of zeroes into files sometimes. I'm changing the component to knfsd. Supplying
a simple, reproducible test case will probably expedite a solution, but don't
be surprised if the problem cannot be reproduced in a different environment.

Comment 2 Alan Cox 2000-02-17 15:45:59 UTC

Please apply the sun patchkits to the machine and see if you can
still reproduce the problem. There is a problem when certain
specific patterns of I/O occur and Solaris corrupts the data.

I don't know if that is your sed problem, but its a good first guess

Comment 3 Jeff Johnson 2000-10-07 14:28:33 UTC

Closed for lack of input.