1 files changed, 84 insertions, 0 deletions
diff --git a/programs/barf/barf.8 b/programs/barf/barf.8
new file mode 100644
index 000000000..e692a4e5f
--- /dev/null
+++ b/programs/barf/barf.8
@@ -0,0 +1,84 @@
+.TH IPSEC_BARF 8 "17 March 2002"
+.\" RCSID $Id: barf.8,v 1.1 2004/03/15 20:35:27 as Exp $
+.SH NAME
+ipsec barf \- spew out collected IPsec debugging information
+.SH SYNOPSIS
+.B ipsec
+.B barf
+[
+.B \-\-short
+]
+.sp
+.SH DESCRIPTION
+.I Barf
+outputs (on standard output) a collection of debugging information
+(contents of files, selections from logs, etc.)
+related to the IPsec encryption/authentication system.
+It is primarily a convenience for remote debugging,
+a single command which packages up (and labels) all information
+that might be relevant to diagnosing a problem in IPsec.
+.PP
+.PP
+The
+.B \-\-short
+option limits the length of
+the log portion of
+.IR barf 's
+output, which can otherwise be extremely voluminous
+if debug logging is turned on.
+.PP
+.I Barf
+censors its output,
+replacing keys
+and secrets with brief checksums to avoid revealing sensitive information.
+.PP
+Beware that the output of both commands is aimed at humans,
+not programs,
+and the output format is subject to change without warning.
+.PP
+.I Barf
+has to figure out which files in
+.I /var/log
+contain the IPsec log messages.
+It looks for KLIPS and general log messages first in
+.IR messages
+and
+.IR syslog ,
+and for Pluto messages first in
+.IR secure ,
+.IR auth.log ,
+and
+.IR debug .
+In both cases,
+if it does not find what it is looking for in one of those ``likely'' places,
+it will resort to a brute-force search of most (non-compressed) files in
+.IR /var/log .
+.SH FILES
+.nf
+/proc/net/*
+/var/log/*
+/etc/ipsec.conf
+/etc/ipsec.secrets
+.fi
+.SH HISTORY
+Written for the Linux FreeS/WAN project
+<http://www.freeswan.org>
+by Henry Spencer.
+.SH BUGS
+.I Barf
+uses heuristics to try to pick relevant material out of the logs,
+and relevant messages
+which are not labelled with any of the tags that
+.I barf
+looks for will be lost.
+We think we've eliminated the last such case, but one never knows...
+.PP
+Finding
+.I updown
+scripts (so they can be included in output) is, in general, difficult.
+.I Barf
+uses a very simple heuristic that is easily fooled.
+.PP
+The brute-force search for the right log files can get expensive on
+systems with a lot of clutter in
+.IR /var/log .