Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 10 Jan 2004 19:25:22 -0500 (EST)
From:      Ted Wisniewski <ted@ness.plymouth.edu>
To:        freebsd-questions@freebsd.org
Subject:   5.2-RELEASE - Show stopper problem
Message-ID:  <200401110025.i0B0PNxL002500@ness.plymouth.edu>

next in thread | raw e-mail | index | archive | help
Here is a description of the problem...

	In 5.2-RC and 5.2-RELEASE there appears to be some issue with
filesystem or I/O subsystem under 5.2-X.   Now, You can install and do the normal
kind of things, however, when you create a lot of I/O on the disk there
seems to be a problem actually reading/writing it to/from disk.   For example,
If I do a "make buildworld"...  It appears to go along ok.   However, I have
had a number of (repeatable) situations where the "make installworld" will go
so far then will not be able to complete.  In this case, there is an attempt
to write data to disk that cannot complete; the process goes into a disk wait 
state (it cannot be killed, and will stay in this state ...  forever).

For example the standard daily security script:

  727  p0  T      0:00.00 sh 100.chksetuid
  737  p0  T      0:02.14 find /usr -xdev -type f ( -perm -u+x -or -perm -g+x -or -perm -o+x ) ( -perm -u+s -or -perm -g+s


I have experienced the same problem on a number of hardware types (Servers like 
New Dell Power edge 6550's and variety of PC's).  For some reason, it appears
that data cannot be read or written to/from the disk.   Possibly due to some
lock never being released.  So, the only way to get rid of the process is to 
reboot...  During the reboot, the system complains about not being able to flush
buffers and "gives up" and on boot, all the filesystems are "dirty" and must be 
FSCK'ed.  Originally, I thought it had something to do with the server model,
but it does occur on a number of Desktop model (a variety of vendors) machines 
as well.  I did send in a PR or two a couple of weeks ago)...  but since I see 
RELEASE tag...  Anyway, If someone could shed any light on why...  I have not been
able to repeat this behavior on 4.9.  

Ted



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200401110025.i0B0PNxL002500>