On Fri, Sep 12, 2003 at 06:28:32AM -0700, Klingensmith, Rick wrote:
> I'm still not having any luck sorting this problem out. I do have a linux
> box and home, but I haven't done a lot with it and will need to install
> SWISH-e on it so the learning curve may be a little steep since I live in
> the windows world too much.

Post if you need help.  Getting a backtrace off linux is not hard.  Just 
make sure you are indexing the exact same data in both tests -- and try 
to see if you can find the smallest data set that shows the problem.

> I checked the web pages before and after the error message from the last
> successful run and they seem OK, nothing out of the ordinary that I can see
> in the source code.
> 
> Any ideas on what would drive SWISH-E to a null pointer?

Not without a backtrace to find out where it's happening.  If you can 
index with pr1 without problems but pr3 aborts, then it's a matter of 
looking at the cvs diffs, but it would be much easier to know where to 
look.


> 
> Rick
> 
> 
> > 
> > On Thu, Sep 11, 2003 at 08:33:19AM -0700, Klingensmith, Rick wrote:
> > 
> > > I am receiving the following warning in my log files from the indexing
> > job:
> > > Warning: Configuration setting for TmpDir 'D:/Inetpub/Indexes/Temp' will
> > be
> > > overridden by environment setting 'C:\DOCUME~1\rek\LOCALS~1\Temp' which
> > does
> > > not exist. When I look in the specified temp directory I've found SWISH-
> > e
> > > work files so I'm not sure if this is a problem or not.
> > 
> > I think swish-e is giving that warning because you are using TmpDir but
> > it found a setting in your environment for TMPDIR, TMP, or TEMP.
> > 
> I thought of changing this for the user context this process is running
> under, but was not sure if it is part of the potential problem or what else
> could go wrong.
> > 
> > > 1468 files indexed.  39,839,610 total bytes.  810188 total words.
> > > Elapsed time: 00:32:05 CPU time: 00:32:05
> > 
> > I wonder why your CPU time is the same as Elapsed time.  32 minutes of
> > CPU is a lot for 1468 files, I'd think.  Maybe that's a Windows thing.
> > 
> 
> This may be just a windows thing, but I run the process at night when the
> machine sees little use, but it should also be dedicating CPU time to the
> web server since it is on the same box.
> 
> > > The archives and FAQ point to the -e option to fix memory issues. What
> > have
> > > I missed?
> > [...]
> > > We've received an error while indexing our website the last two
> > > attempts. The error reads The instruction at "0x0041cca7" referenced
> > > memory at "0x00000000". The memory could not be "read".
> > 
> > Well, I don't know windows error messages, but it looks like something
> > tried to read a null pointer.
> > 
> > Do you have a Linux machine on your network?  Can you try indexing the
> > same files there?
> > 
> Unfortunately I don't have a Linux box to test with. Any other ideas from
> the windows users of the group.
> 
> > --
> > Bill Moseley
> > moseley@hank.org
> 
> Rick Klingensmith
> MSUHuman Resources
> 

-- 
Bill Moseley
moseley@hank.org