Subject: kern/13602: something not kosher in 1.5.1 nfs + ray
To: None <gnats-bugs@gnats.netbsd.org>
From: Tim Rightnour <root@polaris.garbled.net>
List: netbsd-bugs
Date: 07/30/2001 14:52:05
>Number:         13602
>Category:       kern
>Synopsis:       something not kosher in 1.5.1 nfs + ray
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Jul 30 14:47:00 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     Tim Rightnour
>Release:        NetBSD 1.5.1
>Organization:
	
>Environment:
	
System: NetBSD polaris 1.5.1 NetBSD 1.5.1 (POLARIS) #1: Fri Jul 7 16:47:17 MST 2000 root@polaris:/usr/src/1.5.1/sys/arch/i386/compile/POLARIS i386


>Description:
When copying a file from my ray equipped laptop to my NFS server via a mounted
drive, The file copies right up until the last "chunk" and then hangs.

tcpdump shows:

14:42:30.667960 cursa.160561834 > polaris.nfs: 1472 write [|nfs] (frag 17218:1480@0+)
14:42:30.686961 cursa > polaris: (frag 17218:1480@1480+)
14:42:30.700834 cursa > polaris: (frag 17218:1480@2960+)
14:42:30.714679 cursa > polaris: (frag 17218:1480@4440+)
14:42:30.728392 cursa > polaris: (frag 17218:1480@5920+)
14:42:30.742167 cursa > polaris: (frag 17218:1480@7400+)
14:42:30.756702 cursa > polaris: (frag 17218:1480@8880+)
14:42:30.768804 truncated-ip - 1139 bytes missing!cursa > polaris: (frag 17218:1480@10360+)
14:42:30.783577 cursa > polaris: (frag 17218:1480@11840+)
14:42:30.797416 cursa > polaris: (frag 17218:1480@13320+)

This repeats ad-infinitum.

During this time.. the laptop is still responsive to pings.  Only NFS is
affected.

Interestingly.. this only happens when copying files.  These files are 1mb jpg
images.  cd'ing to my pkgsrc tree (mounted via NFS from the same server) and
issuing a make install of a number of large packages never causes this. Normal
usage of the machine such as netscape and whatnot works perfectly as well.
When the command to copy this file however is executed.. the file copies
up to a certain point.. and then just goes to hell.  This is repeatable, and
happens every time I try to copy one of these files.

Reverting my kernel to 1.5_ALPHA2 makes the problem dissapear completely.
	
>How-To-Repeat:
Mount via NFS a directory off a 1.5.1 server over a ray link.  Copy a file.
Lose. I'm not sure if this is just the ray, or just NFS.
	
>Fix:
ENOCLUE
	
>Release-Note:
>Audit-Trail:
>Unformatted: