Date: Mon, 10 Dec 2001 04:20:34 -0800 (PST) From: Mark Hannon <markhannon@optushome.com.au> To: freebsd-gnats-submit@FreeBSD.org Subject: kern/32668: NFS directory removal problems manifested in portupgrade Message-ID: <200112101220.fBACKYT15330@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 32668
>Category: kern
>Synopsis: NFS directory removal problems manifested in portupgrade
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: freebsd-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Mon Dec 10 04:30:01 PST 2001
>Closed-Date:
>Last-Modified:
>Originator: Mark Hannon
>Release: 4.4-STABLE (two versions ... see below)
>Organization:
>Environment:
NFS client:
tbird:~# uname -a
FreeBSD tbird.home.lan 4.4-STABLE FreeBSD 4.4-STABLE #0: Sat Dec 1 22:39:55 EST 2001 mark@tbird.home.lan:/tmp i386
NFS server:
doorway:~> uname -a
FreeBSD doorway.home.lan 4.4-STABLE FreeBSD 4.4-STABLE #0: Sun Oct 7 12:59:46 EST 2001 mark@tbird-fe1.home.lan:/site/arch/FreeBSD-i386/usr/src/sys/compile/DOORWAY_IPFW i386
>Description:
I run a NFS distributed /usr/{compat,local,X11R6} as well as /var/db/pkg
and /etc/X11 small network. Recently I have noticed some problems when
running portupgrade from the NFS client. Upgrades, package removals
etc. all fail with an error message when attempting to delete the old
pkg database directory. The error message is directory not empty.
Running the command from the NFS server works fine.
A ktrace looks ok. The interesting bits are below:
65158 rm RET break 0
65158 rm CALL lstat(0x807d140,0x807d160)
65158 rm NAMI "/var/db/pkg/rmoldmail-0.1.0"
65158 rm RET lstat 0
65158 rm CALL open(0x806bed0,0,0)
65158 rm NAMI "."
65158 rm RET open 3
65158 rm CALL fchdir(0x3)
65158 rm RET fchdir 0
65158 rm CALL access(0x807c000,0x2)
65158 rm NAMI "/var/db/pkg/rmoldmail-0.1.0"
65158 rm RET access 0
65158 rm CALL stat(0x807c000,0xbfbff368)
65158 rm NAMI "/var/db/pkg/rmoldmail-0.1.0"
65158 rm RET stat 0
65158 rm CALL open(0x807c000,0x4,0)
65158 rm NAMI "/var/db/pkg/rmoldmail-0.1.0"
65158 rm RET open 6
65158 rm CALL fstat(0x6,0xbfbff368)
65158 rm RET fstat 0
65158 rm CALL fcntl(0x6,0x2,0x1)
65158 rm RET fcntl 0
65158 rm CALL __sysctl(0xbfbff220,0x2,0x80772f0,0xbfbff21c,0,0)
65158 rm RET __sysctl 0
65158 rm CALL fstatfs(0x6,0xbfbff268)
65158 rm RET fstatfs 0
65158 rm CALL break(0x807f000)
65158 rm RET break 0
65158 rm CALL fstat(0x6,0xbfbff368)
65158 rm RET fstat 0
65158 rm CALL fchdir(0x6)
65158 rm RET fchdir 0
65158 rm CALL getdirentries(0x6,0x807e000,0x1000,0x807a0b4)
65158 rm RET getdirentries 512/0x200
65158 rm CALL lstat(0x807d240,0x807d24c)
65158 rm NAMI "+CONTENTS"
65158 rm RET lstat 0
65158 rm CALL lstat(0x807d340,0x807d348)
65158 rm NAMI "+DESC"
65158 rm RET lstat 0
65158 rm CALL lstat(0x807d440,0x807d44c)
65158 rm NAMI "+COMMENT"
65158 rm RET lstat 0
65158 rm CALL getdirentries(0x6,0x807e000,0x1000,0x807a0b4)
65158 rm RET getdirentries 0
65158 rm CALL lseek(0x6,0,0,0,0)
65158 rm RET lseek 0
65158 rm CALL close(0x6)
65158 rm RET close 0
65158 rm CALL access(0x807d240,0x2)
65158 rm NAMI "+CONTENTS"
65158 rm RET access 0
65158 rm CALL unlink(0x807d240)
65158 rm NAMI "+CONTENTS"
65158 rm RET unlink 0
65158 rm CALL access(0x807d340,0x2)
65158 rm NAMI "+DESC"
65158 rm RET access 0
65158 rm CALL unlink(0x807d340)
65158 rm NAMI "+DESC"
65158 rm RET unlink 0
65158 rm CALL access(0x807d440,0x2)
65158 rm NAMI "+COMMENT"
65158 rm RET access 0
65158 rm CALL unlink(0x807d440)
65158 rm NAMI "+COMMENT"
65158 rm RET unlink 0
65158 rm CALL fchdir(0x3)
65158 rm RET fchdir 0
65158 rm CALL rmdir(0x807c000)
65158 rm NAMI "/var/db/pkg/rmoldmail-0.1.0"
65158 rm RET rmdir -1 errno 66 Directory not empty
65158 rm CALL write(0x2,0xbfbfed80,0x4)
65158 rm GIO fd 2 wrote 4 bytes
"rm: "
65158 rm RET write 4
65158 rm CALL write(0x2,0xbfbfeda0,0x1b)
65158 rm GIO fd 2 wrote 27 bytes
"/var/db/pkg/rmoldmail-0.1.0"
My reading of the above says that all of the unlinks succeed but
the directory is not empty anyway ... perhaps a timing bug with
one of those .nfs files ??
The NFS server has had softupdates both enabled and disabled with no
change in behaviour
>How-To-Repeat:
>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted:
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-bugs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200112101220.fBACKYT15330>