Received: (qmail 2560 invoked by uid 2012); 26 Oct 1997 20:11:36 -0000
Message-Id: <19971026201136.2559.qmail@hyperreal.org>
Date: 26 Oct 1997 20:11:36 -0000
From: Mitch Strickland <strickm@strickland.com>
Reply-To: strickm@strickland.com
To: apbugs@hyperreal.org
Subject: System Panics after running fsck and re-hup-ing or stopping Apache process
X-Send-Pr-Version: 3.2

>Number:         1322
>Category:       mod_status
>Synopsis:       System Panics after running fsck and re-hup-ing or stopping Apache process
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    apache
>State:          closed
>Class:          mistaken
>Submitter-Id:   apache
>Arrival-Date:   Sun Oct 26 12:20:00 PST 1997
>Last-Modified:  Sun Oct 26 12:30:01 PST 1997
>Originator:     strickm@strickland.com
>Organization:
>Release:        1.2.4
>Environment:
SunOS tango 5.5 Generic_103093-14 sun4m sparc SUNW,SPARCstation-20, Apache 1.2.4 with mod_status compiled
>Description:
System Panics after running fsck, clearing an unreferenced file stored in the logs directory, and re-hup-ing or stopping Apache process.
>How-To-Repeat:
1. Run fsck on directory containing Apache server root directory on Solaris 2.x
2. Clear UNREF file with inode number consistent with inode numbers is Apache's logs directory
3. Continue to run fsck until no problems reported
4. HUP or stop Apache process
5. 90 - 100% of the time system will panic with message "freeing the free inode"
>Fix:
It appears the UNREF file is associated with the 'scoreboard' file. In a future release find a way to create this file so that it does not allow fsck to think their is a problem with the filesystem
>Audit-Trail:
State-Changed-From-To: open-closed
State-Changed-By: coar
State-Changed-When: Sun Oct 26 12:28:38 PST 1997
State-Changed-Why:
The problem is that you're trying to repair a disk while
it's in use, which is a major badness under any circumstances.
If you try to change a tyre while a vehicle is in motion,
and you get run over, you've no-one but yourself to blame.

Apache opens some files and immediately unlinks them to
prevent any other process from being able to access (and
particularly modify) them.  This is fairly common
practice.

The workaround, or actually the correct procedure, is
for you to do your repairs on a quiescent filesystem.

Thanks for using Apache!
Class-Changed-From-To: sw-bug-mistaken
Class-Changed-By: coar
Class-Changed-When: Sun Oct 26 12:28:38 PST 1997
Severity-Changed-From-To: critical-non-critical
Severity-Changed-By: coar
Severity-Changed-When: Sun Oct 26 12:28:38 PST 1997

From: Dean Gaudet <dgaudet@arctic.org>
To: Mitch Strickland <strickm@strickland.com>
Cc: apbugs@hyperreal.org
Subject: Re: mod_status/1322: System Panics after running fsck and re-hup-ing or stopping Apache process
Date: Sun, 26 Oct 1997 12:32:25 -0800 (PST)

 Are you running fsck on a filesystem which is live, being used by Apache?
 
 Dean
 
>Unformatted: