httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rob Hartill <r...@imdb.com>
Subject Re: WWW Form Bug Report: "Configure fails on OSF/1 V2.0" on (fwd)
Date Wed, 04 Dec 1996 21:54:09 GMT
----- Forwarded message from Sftwr Dvlp -----

Date: Wed, 4 Dec 1996 15:26:00 -0500
Message-Id: <9612042026.AA01810@alpha.vaxxine.com>
X-Sender: develop@vaxxine.com
X-Mailer: Windows Eudora Light Version 1.5.2
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
To: Paul Richards <p.richards@elsevier.co.uk>
From: Sftwr Dvlp <develop@vaxxine.com>
Subject: Re: WWW Form Bug Report: "Configure fails on OSF/1 V2.0" on
  OSF/1
Cc: robh@imdb.com

At 05:58 PM 12/4/96 +0000, you wrote:
>Rob Hartill <robh@imdb.com> writes:
>
>> Hmm, very strange.
>> 
>> I'll forward the info to the developers list.
>
>> >Symptoms:
>> >--
>> >The following results from an attempt to run
>> >Configure.  I've tried it on two OSF/1 boxes with
>> >lots o' available memory.
>> >
>> >Using 'Configuration' as config file
>> > + configured for DEC OSF/1 platform
>> >swap space below 10 percent free
>> >Unable to obtain requested swap space
>> >Unable to obtain requested swap space
>
>A stab in the dark. Is this person using a swap file mounted on an
>ordinary filesystem which is almost full?
>
>Some VM systems allocate swap to pages even if there's lots of
>free memory. If OSF is such a system and he's also making use of an
>additional swap file mounted on a ufs partition then that might cause
>the above to happen. If he wants help debugging this then the output
>of df and swapinfo (or the OSF equivalent) would be the first step.
>
>In any case, it's *extremely* unlikely that this is an Apache bug.
>

No additional swap file here.  It's a good point though and might explain
some of my problems with other software. :)

In the case of Apache, however, it seems that there may be yet another
oddity in our OSF/1 2.0.  I tracked down the cause of my problem to the
"PrintPath" helper script.  I've included a version that works on both
of the alphas I tested on.  An additional nicety would be if the cc flag
-std1 was used on OSF/1 since at least one source file uses ANSI style
#if directives which OSF/1 wants this flag for.

-------- PrintPath --------

#!/bin/sh
# Look for $1 somewhere in $PATH
#  will print out the full pathname unless
#  called with the '-s' option
#

if [ $1 = "-s" ]; then
    shift
    echo="#"
else
    echo="echo"
fi

for path in `echo $PATH |
 sed 's/^:/.:/
      s/::/:.:/g
      s/:$/:./
      s/:/ /g' `
do
    if [ -x $path/$1 ]; then
        # Changed below from attempting to "eval # x/y" due to an apparent
        # problem in OSF/1 that would cause sh to grow to consume all
        # available memory when evaluating the resulting comment.  On a
        # test alpha the size of sh grew to about 181 MB according to "top".
        if [ "$echo" != "#" ]; then
      	  eval $echo $path/$1
        fi
	exit 0
    fi
done
exit 1
--
Rob Kreamer                                *  develop@vaxxine.com
(Vaxxinated) Software Development Manager  *  http://vaxxine.com/kreamer
--
Internet Service From the Friendly Folks at
Vaxxine Computer Systems <admin@vaxxine.com> (905) 562-3500
Niagara Region, Ontario, CANADA


----- End of forwarded message from Sftwr Dvlp -----


Mime
View raw message