Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 3 Jun 1997 19:18:40 -0700 (PDT)
From:      John Sin <jsin@broadcom.com>
To:        FreeBSD-gnats-submit@FreeBSD.ORG
Subject:   kern/3771: bug report for 2.2.2-RELEASE
Message-ID:  <199706040218.TAA02578@airbus.broadcom.com>
Resent-Message-ID: <199706040220.TAA11237@hub.freebsd.org>

next in thread | raw e-mail | index | archive | help


>Number:         3771
>Category:       kern
>Synopsis:       NFS hangs when writing to local FS re-mounted via NFS
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    freebsd-bugs
>State:          open
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Jun  3 19:20:01 PDT 1997
>Last-Modified:
>Originator:     John Sin
>Organization:
Broadcom Corporation
>Release:        FreeBSD 2.2.2-RELEASE i386
>Environment:

	2.2.2-RELEASE FreeBSD 2.2.2-RELEASE #0: Sat May 31 15:24:00 PDT 1997     

	DELL Pentium Pro Edge 2100/180.  64MB RAM.  
	On-Board AIC7880 SCSI plus Adaptec 2940 UW.  3COM 3c595 100BaseTX.
	3 SCSI hard drives and SCSI CDROM.

	DELL Pentium II H266.  96MB RAM.  
	On-Board EIDE controller.  3COM 3c905 100BaseTX.
	1 IDE hard drive and an IDE CDROM


>Description:

	NFS hangs, when writing a sizable file to a local filesystem 
	that has been re-mounted via NFS (typically via amd, but not
	necessarily).

	The reason for re-mounting local filesystem via NFS is so
	that directory names can be kept the same across all machines.

	Small files can be written ok to such a filesystem (it seems).  
	Large files can be read OK.  Whenever the system dies, all NFS
	daemons are in 'D' state, with very high priority.  Here's
	a sample of "ps axl | grep nfs":

UID   PID  PPID CPU PRI NI   VSZ  RSS WCHAN  STAT  TT       TIME COMMAND
  0    96     1   9   2  0   232  116 accept Is    ??    0:00.01 nfsd: master
  0   100    96   0 -14  0   224  104 uihget D     ??    0:00.51 nfsd: server
  0   101    96   0  -1  0   224  104 nfsrcv D     ??    0:00.05 nfsd: server
  0   102    96   0 -14  0   224  104 uihget D     ??    0:00.01 nfsd: server
  0   103    96   0 -14  0   224  104 uihget D     ??    0:00.00 nfsd: server
  0   109     1   0   2  0   208   88 sbwait D     ??    0:00.24 nfsiod -n 4
  0   110     1   0  -1  0   208   88 nfsrvc D     ??    0:00.06 nfsiod -n 4
  0   111     1   0  -1  0   208   88 nfsrvc D     ??    0:00.01 nfsiod -n 4
  0   112     1   0  -1  0   208   88 nfsrvc D     ??    0:00.01 nfsiod -n 4
100   385   170   0  -1  0   260  156 nfsrvc D+    ??    0:00.14 cp /tmp/file

	It does appear that NFS in general works -- at least I can mount 
	remote machines' filesystems and copy data in and out without the 
	system hanging.  I only experience NFS problem with local filesystem
	re-mounted via NFS.

>How-To-Repeat:

	cat /kernel >  /tmp/junkfile
	cat /kernel >> /tmp/junkfile
	cat /kernel >> /tmp/junkfile
	[hopefully this results in /tmp/junkfile > 3MB]

	mount localhost:/ mnt
	cp /mnt/tmp/junkfile /tmp/junkfile.copy		[this works ok
	cp /tmp/junkfile /mnt/tmp/junkfile.dead		[this hangs the system]

>Fix:
	
	Unknown

>Audit-Trail:
>Unformatted:



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199706040218.TAA02578>