Age | Commit message (Collapse) | Author |
|
The CPU checks were fireing way too often. Bump up the multipliers to
reduce the number of alerts.
|
|
The "check_load" nagios check should not be hardcoded to some arbitrary
value. The thresholds should rather be adjusted in function of how many
CPUs are present in the server.
The factors were chosen by prior experience but could be adjusted if
needed.
Factors for warning thresholds (where N is the nb of cpus):
* 1min : 5 * N
* 5mins : 4 * N
* 15mins : 3 * N
Factors for critical thresholds:
* 1min : 10 * N
* 5mins : 9 * N
* 15mins : 8 * N
|
|
By default, nrpe listens on every interfaces on the server. This is
undesirable on Linux VServer hosts, since it will conflict with
vservers.
With a new variable, $nagios_nrpe_server_address, it is now possible to
set the ip address on which the nrpe server will listen.
|
|
This package doesn't exist on lenny.
|
|
Make some changes to ensure compatibility of the nrpe resources with
FreeBSD.
* Use group 0 instead of 'root'
* capitalize values for testing $operatingsystem
* override arguments for the nrpe service so that FreeBSD can find the
init script
|
|
The 'command' resource defines files under the root group. Since the
'root' group does not exist by default on FreeBSD, make an exception for
FreeBSD to use the 'wheel' group instead.
|
|
Configuration needs to be placed in a different directory in FreeBSD.
Some packages are not named the same under FreeBSD.
Some packages are not available under FreeBSD. Migrate those to a new
nagios::nrpe::linux class so that FreeBSD does not take them into
account.
|
|
We need to be able to configure what NRPE commands are defined on
clients.
This new method accepts inline contents or a puppet source argument.
Signed-off-by: Gabriel Filion <lelutin@gmail.com>
|
|
amended (lelutin): took out some trailing whitespaces and a commented
out (useless) line
|
|
|
|
|
|
|
|
|
|
amended (lelutin): removed trailing spaces and commented out code.
|
|
The default nsca.cfg is currently using the "non-normal" port for nsca.
Let's import a freshly installed nsca.cfg file with default values.
Signed-off-by: Gabriel Filion <lelutin@gmail.com>
|
|
Without this link, puppet sometimes attempts to start the service before
the package is installed, wich triggers an error.
Signed-off-by: Gabriel Filion <lelutin@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Conflicts:
templates/irc_bot/nsa.cfg.erb
|
|
|
|
|
|
|
|
|
|
Fully qualify facter variables to suppress warnings
|
|
|
|
|
|
The current code tries to get the content of those two files in a bunch
of places and defines a fallback to a generic file in the module's
files/configs directory.
The problem is that there is no such generic config file present.
Link to Debian's config files as a generic version of those
configuration files.
This bug was found by taggart.
|
|
|
|
|
|
This reverts commit fb75fcec220f9e2943d2e7af292ff884fd237f6d.
Reverted due to LeLutin's comments in #3310 and the fix implemented in #3298 in
preparation of merging those
|
|
|
|
|
|
Conflicts:
manifests/irc_bot.pp
Had to merge in the use_notices change that we had in our branch
|
|
|
|
This gives us the possibility of using the nagios::irc_bot::disable
class without having any config variable set. When using
nagios::irc_bot, we still get yelled at when we didn't define the
appropriate variables.
While we're moving the code around, change the group applied on the
different files from 'root' to '0' for better portability.
|
|
All classes serving the same purpose as this one are called 'disable',
so let's follow this standard.
|
|
A misconception on my part made the dependencies "not-quite-right" (as
in: the script doesn't really depend on nagios, as we would expect). This
gives the IRC bot the save "level" as nagios itself. However, we'd like
for nagios to be started before we start the IRC bot.
|
|
It may sometimes happen that the bot dies or gets killed and the socket
file is not cleaned up. When using the init script with 'start', as
puppet does, the socket file blocks the bot from starting.
|
|
Since the service "disable" bug was fixed in the debian service provider
[1] in the puppet 2.6.2-5 package in squeeze, let's switch from the
manual calls to the "enable" feature.
[1]: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=573551
Also add a requirement between service[nagios-nsa] and
file[/etc/init.d/nagios-nsa]. This was ensured in Debian with the exec
that we're now replacing but didn't find its way through the
centosification.
|
|
For when you need to ensure that the bot doesn't start on a server.
|
|
When we modify the config file, we'd like for the service to restart in
order for it to use the new configuration.
|
|
To give a little more info/meaning on what the number $SERVICEATTEMPT$ is,
change the current notification from "... n.$SERVICEATTEMPT$" to "...
$SERVICEATTEMPT$/$MAXSERVICEATTEMPTS$".
|
|
|
|
|
|
|