Provided by:
hobbit_4.2.0.dfsg-10_i386 
NAME
client-local.cfg - Local configuration settings for Hobbit clients
SYNOPSIS
~hobbit/server/etc/client-local.cfg
DESCRIPTION
The client-local.cfg file contains settings that are used by each
Hobbit client when it runs on a monitored host. It provides a
convenient way of configuring clients from a central location without
having to setup special configuration maintenance tools on all clients.
The client-local.cfg file is currently used to configure what logfiles
the client should fetch data from, to be used as the basis for the
"msgs" status column; and to configure which files and directories are
being monitored in the "files" status column.
Note that there is a dependency between the client-local.cfg file and
the hobbit-clients.cfg(5) file. When monitoring e.g. a logfile, you
must first enter it into the client-local.cfg file, to trigger the
Hobbit client into reporting any data about the logfile. Next, you must
configure hobbit-clients.cfg so the Hobbit server knows what to look
for in the file data sent by the client. So: client-local.cfg defines
what raw data is collected by the client, and hobbit-clients.cfg
defines how to analyze them.
PROPAGATION TO CLIENTS
The client-local.cfg file resides on the Hobbit server.
When clients connect to the Hobbit server to send in their client data,
they will receive part of this file back from the Hobbit server. The
configuration received by the client is then used the next time the
client runs.
This method of propagating the configuration means that there is a
delay of up to two poll cycles (i.e. 5-10 minutes) from a configuration
change is entered into the client-local.cfg file, and until you see the
result in the status messages reported by the client.
FILE FORMAT
The file is divided into sections, delimited by "[name]" lines. A
section name can be either an operating system identifier - linux,
solaris, hp-ux, aix, freebsd, openbsd, netbsd, darwin - or a hostname.
When deciding which section to send to a client, Hobbit will first look
for a section named after the hostname of the client; if such a section
does not exist, it will look for a section named by the operating
system of the client. So you can configure special configurations for
individual hosts, and have a default configuration for all other hosts
of a certain type.
Apart from the section delimiter, the file format is free-form, or
rather it is defined by the tools that make use of the configuration.
LOGFILE CONFIGURATION ENTRIES
A logfile configuration entry looks like this:
log:/var/log/messages:10240
ignore MARK
trigger Oops
The log:FILENAME:SIZE line defines the filename of the log, and the
maximum amount of data (in bytes) to send to the Hobbit server.
FILENAME is usually an explicit full-path filename on the client. If it
is enclosed in backticks, it is a command which the Hobbit client runs
and each line of output from this command is then used as a filename.
This allows scripting which files to monitor, e.g. if you have logfiles
that are named with some sort of timestamp.
The ignore PATTERN line (optional) defines lines in the logfile which
are ignored entirely, i.e. they are stripped from the logfile data
before sending it to the Hobbit server. It is used to remove completely
unwanted "noise" entries from the logdata processed by Hobbit.
"PATTERN" is a regular expression.
The trigger PATTERN line (optional) is used only when there is more
data in the log than the maximum size set in the "log:FILENAME:SIZE"
line. The "trigger" pattern is then used to find particularly
interesting lines in the logfile - these will always be sent to the
Hobbit server. After picking out the "trigger" lines, any remaining
space up to the maximum size is filled in with the most recent entries
from the logfile. "PATTERN" is a regular expression.
COUNTING LOGENTRIES
A special type of log-handling is possible, where the number of lines
matching a regular expressions are merely counted. This is
linecount:FILENAME, followed by a number of lines of the form
ID:PATTERN. E.g.
linecount:/var/log/messages
diskerrors:I/O error.*device.*hd
badlogins:Failed login
FILE CONFIGURATION ENTRIES
A file monitoring entry is used to watch the meta-data of a file:
Owner, group, size, permissions, checksum etc. It looks like this:
file:/var/log/messages[:HASH]
The file:FILENAME line defines the filename of the file to monitor. As
with the "log:" entries, a filename enclosed in backticks means a
command which will generate the filenames dynamically. The optional
[:HASH] setting defines what type of hash to compute for the file: md5,
sha1 or rmd160. By default, no hash is calculated.
NOTE: If you want to check multiple files using a wildcard, you must
use a command to generate the filenames. Putting wildcards directly
into the file: entry will not work.
DIRECTORY CONFIGURATION ENTRIES
A directory monitoring entry is used to watch the size of a directory
and any sub-directories. It looks like this:
dir:DIRECTORYNAME
The dir:DIRECTORYNAME line defines the filename of the file to monitor.
As with the "log:" entries, a filename enclosed in backticks means a
command which will generate the filenames dynamically. The Hobbit
client will run the du(1) command with the directoryname as parameter,
and send the output back to the Hobbit server.
NOTE: If you want to check multiple directories using a wildcard, you
must use a command to generate the directory names. Putting wildcards
directly into the dir: entry will not work. E.g. use something like
dir:‘find /var/log -maxdepth 1 -type d‘
The "du" command used can be configured through the DU environment
variable. On some systems, by default du reports data in disk blocks
instead of KB (e.g. Solaris). So you may want to configure the Hobbit
client to use a du command which reports data in KB, e.g. by setting
DU="du -k"
in the hobbitclient.cfg file.
NOTES
The ability of the Hobbit client to calculate file hashes and monitor
those can be used for file integrity validation on a small scale.
However, there is a significant processing overhead in calculating
these every time the Hobbit client runs, so this should not be
considered a replacement for host-based intrusion detection systems
such as Tripwire or AIDE.
Use of the directory monitoring on directory structures with a large
number of files and/or sub-directories can be quite ressource-
intensive.
SEE ALSO
hobbit-clients.cfg(5), hobbitd_client(8), hobbitd(8), hobbit(7)