Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 14 Dec 1996 10:52:39 -0500 (EST)
From:      John Fieber <jfieber@indiana.edu>
To:        Ade Barkah <mbarkah@hemi.com>
Cc:        jkh@time.cdrom.com, www@freebsd.org
Subject:   Re: Follow-up to FreeBSD documentation team status report.
Message-ID:  <Pine.BSI.3.95.961214102452.7340d-100000@fallout.campusview.indiana.edu>
In-Reply-To: <199612140412.VAA23370@hemi.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, 13 Dec 1996, Ade Barkah wrote:

> Easy to change fortunately. How far back are logs usually kept ?

I'd like to keep a month or two on hand but, as you notice, the
rotating is, um, less than sophisticated.  :)

Too bad apache doesn't use syslog, then rotation would be a snap. 
We would just have to sed off the beginnings of lines before
feeding the logs to stats software.  Hm... anyone want to look
into this possibility? 

> Maybe just move the www CVS tree to spatter ? Solves a lot of
> problems.

Anybody have an opinion on this?  Honestly, the thought didn't
even occur to me.  I'm a cvs *user*, not an administrator so I
really can't comment on the implications of this.

> How often are the mailing list archives updated ? db/text shows
> the last update was three days ago. Which program(s) append
> new mail to the archives ?

The mailing lists are updated in a very ad-hoc fashion. 
Basically when I feel moved to do it.  Part of the problem is
negotiating a workable method with the postmaster.  When I
started all this (the mailing lists, which gave birth to the web
site...) the mailing list archives were rotated on a weekly
basis.  Now they are not rotated at all so every update indexes
from scratch.  /usr/local/www/bin/mailindex does the updating.

Yes, waisq crashes on a regular basis, but to date, I have not
been able to personally reproduce this on demand!  Unfortunately
the crash times are not logged so I can't easily correlate the
crashes with the web log files to see what queries were involved.

...Of course search.cgi is probably in position to detect the
crash and log the search...   Of course, now that some people
have offered their services to work with cgi scripts and mailing
list archive....

Those looking into search.cgi, just be warned that it was the
first perl program I ever wrote and it is pretty hideous.
Independent of adding new functionality, it simply needs to be
rewritten for aesthetic reasons.

[And for those who really do look into search.cgi, the reason for
not specifying the database on the waisq command line is that you
can only specify one.  However, if you actually sit down and play
with querying various mailing lists separately, and then
together, you will realize that freewais totally sucks at
searching multiple databases anyway and the feature should simply
be removed.  There are also a number of very basic reasons why
searching multiple databases and ranking the results is doomed to
failure anyway.  I can provide citations to articles discussing
the issue if anyone is interested.]

> Will... you still run a www server on freefall then, to support
> the cgi ? 

Unless all CGIs can be moved, yes.  Even then, it might be a good
idea to leave a server running that simply redirects to spatter.
I expect there are a fair number of bookmarks and web databases
in the world that explicitly reference freefall.  :(

> Or move the cgis over to spatter as well and simply
> have them access nfs-mounted data (cvs/mail archives) from
> freefall ?

That would work so long as the indexes are on spatter.  Indexing
is very disk intensive and probably not a good use of network
bandwidth.  :)   Spatter currently has 1.7GB free on /usr so
unless someone else has plans for that space, I don't see any
problem.

The other CGI to worry about is the query-pr.cgi.  I've never
looked at it so I don't know what the options for moving it are.

> is moved into spatter's ~ncvs(?). I can't seem to find any references
> to webupdate from /etc/crontab or /etc/daily. Is this called from
> someone's crontab right now ?

Mine.  :)

Whoever runs it, the output should probably go to the www list so
people can see when they messed up.  (The update just bails out
if errors are found.)

-john




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSI.3.95.961214102452.7340d-100000>