Hi,
So... no answers... bug report filed! :-)
https://tickets.puppetlabs.com/browse/MODULES-2376
Matthias
On Mon, 3 Aug 2015 13:26:07 +0200
Matthias Saou wrote:
> Hi,
>
> I had this applied to my nodes :
>
> firewall { "${prenumber}7 portknock let connections th
se or even remove the $rseconds line entirely,
puppet just leaves the previous value on existing nodes. For new nodes
or if I manually remove all iptables rules first, then the new rule
gets created without any "--seconds 3" as expected.
How can I tell puppet to actually r
On Wed, 20 May 2015 03:57:50 -0700 (PDT)
Nick Howes wrote:
> On Wednesday, 20 May 2015 11:52:06 UTC+1, Matthias Saou wrote:
> >
> > I will post followups if
> > anyone is interested.
> >
> >
> Count me interested - I'm in the same pre-leap state (3
a quick search doesn't come up with much relevant.
This is a RHEL7 system with the default LANG=en_US.UTF-8...
Not going so well so far, I was really hoping for less differences
between 3.7 with the future parser and 4.0. I will post followups
t to get it as a number instead of a string :
"mtu_eth0": 1500,
"operatingsystemrelease": "6.6",
"swapsize_mb": "4032.00",
"processorcount": 8,
Is this a known limitation? Am I doing something wrong?
Matthias
--
Matthias
On Wed, 27 Aug 2014 05:31:39 -0700 (PDT)
Brian Wilkins wrote:
> I am having to escape a regular expression in my match and puppet
> complains on the puppetmaster of an "Unrecognised escape sequence".
> How do I fix this so the logs don't get cluttered with this message?
> I am using Puppet 3.6.2.
ET" or => { ipv4_10_225_172 };
> "DC2_QA_NET" or => { ipv4_10_224_136 };
>
> For example….
> If your hostname matches the “web_regex” and you reside in the
> DC2_QA_NET then do X, Y and Z.
>
> Thanks in advance for your time and assistance.
>
>
e errors if I check all of the packages have
> been installed. rpm -qa | grep postgres
> postgresql-libs-8.1.23-10.el5_10
> postgresql92-libs-9.2.8-1PGDG.rhel5
> postgresql-libs-8.1.23-10.el5_10
> postgresql92-9.2.8-1PGDG.rhel5
>
> So on the second run everything that was missed,
ing the cache format is added automatically.
> # The default value is '$confdir/localconfig'.
> localconfig = $vardir/localconfig
> certname = <%= fqdn %>
> server = <%= puppetserver %>
> pluginsync = true
> repor
the
> restart (service vs systemctl)?
Take a look at :
http://docs.puppetlabs.com/references/stable/type.html#service
You *should* be able to force the restart command with something like
this (untested) :
service { 'auditd':
restart => '/sbin/service auditd restart'
petlabs.com/puppet/latest/reference/release_notes.html#feature-purging-unmanaged-ssh-authorized-keys
Thank you so so so so so much! Definitely my most wanted feature
since... forever :-) Off to try that now...
Matt
h variable in file source parameters and
template paths to make your life easier.
HTH,
Matthias
--
Matthias Saou ██ ██
██ ██
Web: h
d why one would want to do that... I
could actually use that behaviour for the initial ~/.gitconfig files I
create for system users, for example.
Matthias
--
Matthias Saou ██ ██
██ ██
Web: http://matthias.saou.
.]
> So why has this virtualhost array suddenly become a string
> when I call the each function on it in the template?
Because unless I'm mistaken, you're converting your array to a string
with ".join('", "')". Just re
probably.
Run : /usr/bin/hadoop fs -ls /user/foo; echo $?
If you see 0 then the exec will no longer run when using "unless".
Run : /usr/bin/hadoop fs -mkdir /user/foo; echo $?
You'll need to see 0 when the HDFS directory didn't exis
agios/nagios_hostgroup.cfg',
'/etc/nagios/nagios_service.cfg',
'/etc/nagios/nagios_servicegroup.cfg',
'/etc/nagios/nagios_timeperiod.cfg',
]:
ensure => present,
operatingsystem => RedHat
operatingsystemmajrelease => 7
operatingsystemrelease => 7.0
osfamily => RedHat
rubyversion => 2.0.0
Matthias
--
Matthias Saou ██ ██
██ █
xist :-)
But it's a good start, much appreciated!
Matthias
--
Matthias Saou ██ ██
██ ██
Web: http://matthias.saou.eu/
solved it in my own module :
https://github.com/thias/puppet-nagios/blob/master/manifests/server.pp#L550
(feel free to copy/paste those lines!)
Though I did "move" all of the resources to their default files and
stopped using the objects/templates.cfg file.
HTH,
Matthias
--
; about 15 places left if you want to join in. So, if you want food,
> please register :).
>
> See you soon!
>
--
Matthias Saou ██ ██
██
headers like IfModifiedSince etc.
>
> So here's some instructions on how to do it:
> https://gist.github.com/dalen/8419913
>
> Works for me, but no guarantees etc :)
This is a simple and interesting change. Care to share some numbers
about the "speeds things up quite a bit" par
ACL is
still being applied by the module (remove users easily and
reliably).
* Automatic setting of both normal and default ACLs to the same values
(shortens declarations, increases code readability).
--
Feedback welcome!
Matthias
y separate variables for each role, such as:
webserver: true
appserver: true
With matching facts which exist or don't exist.
HTH,
Matthias
--
Matthias Saou ██ ██
d. I'm guessing the update broke something related to SSL. After
installing 1.7.0, alternatives automatically updated all java related
paths to make 1.7.0 the default, and puppetdb seems to work fine with
it.
So if you're running PuppetDB on RHEL (or any clone), the
Is there some option or
trivial workaround I'm missing?
Matthias
--
Matthias Saou ██ ██
██ ██
Web: http://matthias.saou.eu/ ██
Mai
raring to the default cows list
> e8590a6 (packaging) Remove f16 from mocks to build as it is EOL
> 4b3a085 (packaging) Update FACTERVERSION to 1.7.2-rc1
> 2027595 (packaging) Update FACTERVERSION to 1.7.2
>
> Moses Mendoza (2):
> 2029359 (packaging) Use the pa
::ntp::install'] ->
Class['::ntp::config'] ~> Class['::ntp::service'] ->
Anchor['ntp::end']
Matthias
--
Matthias Saou ██ ██
██ ██
Web: http://mat
"false" -%>
Add a new params variable for it, similar to $panic?
* For "real" RHEL, the ntp server hostnames used will be "centos"
instead of the original "rhel" ones. I'm not su
oo, where do we put the common virtual resource?).
* The use of author-specific common modules (I don't like taking a
johndoe/apache module and noticing I then need johndoe/common).
But don't get me wrong, I like where this is headed, and will
participate as much as I can.
by most. It's what works for many GNU/Linux distributions
(Debian, Fedora, etc.).
Just my 2¢ ;-)
Matthias
--
Matthias Saou ██ ██
██ ██
Web: http://matthias.saou.eu/ ██
Mail/XMPP: matth..
d), something like
this should work, since "-" is unique :
$user-akey = regsubst($akey, '^(.*)$', "${name}-\1")
my_ssh_authorized_key { $user-akey: ensure => $ensure }
Then :
define my_ssh_authorized_key ( $ensure ) {
$user = regsubst
cting me!
Matthias
--
Matthias Saou ██ ██
██ ██
Web: http://matthias.saou.eu/ ██
Mail/XMPP: matth...@saou.eu ██
███
ckage is installed) and found I
> could not do it completely from within the manifest.
>
> “Sometimes I think the surest sign that intelligent life exists
> elsewhere in the universe is that none of it has tried to contact
> us.” Bill Waterson (Calvin & Hobbes)
>
> - Orig
There are other ways. None are nice and clean, but a custom fact just
for this seems overkill.
Here's a quick example of how I've implemented creating a default
~/.gitconfig for users if it doesn't exist, but not modify it if it's
already there or has been modified.
$gitconfig_user_name = $my
repo["puppetlabs-products"]" lines.
Otherwise, you will need to add that resource to be declared.
Matthias
--
Matthias Saou ██ ██
██ ██
Web: http://matthias.saou.eu/
;. I even tried "--no-verbose" but that doesn't work.
How do I switch back to a default log level of "notice"?
Matthias
--
Matthias Saou ██ ██
You could also have a look at this other module :
https://forge.puppetlabs.com/thias/tuned
I just had a look at Romain's, and a provider just for this is
overkill. As much as I hate exec, it's fine here since the profile name
is exactly contained in a single configuration file. Also, a class is
re fine with just ensuring the package is absent, and
enclosing the service inside an "if $ensure == 'present'" or similar.
If that's not enough, it should be considered a packaging bug.
Matthias
--
Matthias Saou ██ ██
. Your explanation matches the behaviour I've
seen.
Matthias
--
Matthias Saou ██ ██
██ ██
Web: http://matthias.saou.eu/ ██
Mail/XMPP: matth
Hi,
You shouldn't be trying with a variable. Here's what you should
probably be doing instead :
/etc/puppet/modules/sensor/manifests/listner.pp
define sensor::listner ( $config = 'this is the generic default' ) {
notify { $config : }
}
/etc/puppet/manifests/site.pp
Sensor::Listener { config
e proper syntax?
Has this (or will this) become the best practice?
Matthias
[1] http://docs.puppetlabs.com/puppet/2.7/reference/lang_classes.html
[2] http://docs.puppetlabs.com/puppet/2.7/reference/lang_scope.html
--
Matthias Saou ██ ██
ur Puppet code, put a line like:
>
> $service_real = str2bool("$service")
>
> I've been calling that pattern "sloppy bools." It will work around
> the bug for now, and continue to work once the bug is fixed and you
> start using real booleans in your y
e default...
Other variable names don't seem to have this problem. It seems that
I've been bitten once more by "mostly-working-but-reserved-somewhere"
variable names.
Does anyone know Hiera internals enough to confirm this?
Do I have any options other than
fully implemented it ? Is there any official
> documentation besides the type reference ? Should I dig in
> thias/nagios ?
>
> Thanks in advance,
>
> Francois
>
>
--
Matthias Saou ██ ██
On Tue, 24 Jul 2012 09:39:30 -0400
Trevor Vaughan wrote:
> I attempted to set it and it wasn't very happy but your assessment is
> what I had thought as well.
Yup. (Still) Unsupported, unfortunately...
See http://projects.puppetlabs.com/issues/1581
Matthias
--
Ma
ance)
This is indeed not Debian, as an update will never get you to the next
major release :-)
EL5 is indeed quite old, EL6 has already been available for nearly 2
years and ships with ruby 1.8.7 : It's definitely what you want to be
deploying on any new systems.
Che
gt; you have Ruby 1.8.7 or greater installed.
--
Matthias Saou ██ ██
██ ██
Web: http://matthias.saou.eu/ ██
Mail/XMPP: matth...@saou.eu ██
On Wed, 25 Apr 2012 10:29:23 -0700 (PDT)
Shantanu wrote:
>
> On Apr 25, 11:46 am, Craig Dunn wrote:
> > On 25/04/2012 17:01, Shantanu wrote:
> >
> >
> >
> > > I would like to 'exec' an installation script 'onlyif' an
> > > installation directory is empty. I tried using following bash
> > > one-
On Thu, 29 Mar 2012 10:06:01 +0200
Martijn Grendelman wrote:
[...]
> >
> > ruby hashes are not stored in predictable order so this will
> > happen, the proposed solution should work.
> >
> > But as always the best is just to test it and see how it goes, it
> > wont bite :)
>
> <% aliases.sort_b
Hi,
Before making any further changes to my existing PHP module, I'd like
to get a few recommendations from anyone who has already faced this
problem...
My existing module supports Fedora, RHEL and clones. I'd like to extend
support to more distributions, by adding a typical "params" class to it.
On Wed, 25 Jan 2012 09:59:10 -0500
Nan Liu wrote:
[...]
> > i.e. Say there were two unrelated modules which said slightly
> > different but *compatible* things:
> >
> > Module a:
> >
> > file { "/foo/bar":
> > ensure => 'present',
> > owner => 'root',
> > content => "blah blah",
> > }
> >
On Tue, 24 Jan 2012 10:32:31 +0100
Felix Frank wrote:
> Perhaps there needs to be some kind of "Forge common" module that by
> policy can only ever declare virtual resources (packages are a
> prominent example).
This only takes care of the Forge case. The problem of having the same
package being
Sans wrote:
> Thanks Matthias! I noticed the semicolon typo.
> What does ${::hostname}-cert.pem implicate (as opposed to ${hostname}-
> cert.pem)?
It just makes explicit that it's a variable from the global scope (as
it's a fact) and not from the current local scope.
Check out recent puppet doc
Matthias Saou wrote :
> file { "/opt/${cert_dir}/${::hostname}-cert.pem":
> mode=> '0444', owner => 'root', group => 'root',
> source => "puppet:///modules/p_nodes/${::hostname}-cert.pem";
>
Sans wrote :
> As said in the subject line, is there a "nice" way of creating
> directory tree if parent is not already there, like using "mkdir -p"?
This will do exactly what you want :
$cert_dir = 'globus/cert'
exec { 'mkdir_cert_dir':
path=> [ '/bin', '/usr/bin' ],
Aaron Grewell wrote :
> Agreed, facts are the one thing that really should be global IMHO.
I'd settle for anything short. If <%= ::factname %> could have worked,
it would have been fine by me.
What's the "official best practice" going to be for facts inside
templates with puppet 2.7+? lookupvar
Corey Osman wrote :
> I need to get the environment variable ORACLE_SID from the OS. I know this
> can be done with the following:
>
> $blah = env("PATH")
>
> However, the ORACLE_SID variable is only set under the oracle user account.
> So I would need a way to login as the oracle account fi
Peter Meier wrote :
> >> Is this a known issue? Is there any work-around? It's really breaking my
> >> CentOS 6 servers' ability to find their REPOs. :-)
> >
> > You need to have the lsb_release command installed for the lsb* results
> > to appear.
> >
> > Fedora/Redhat have it in the redhat-ls
Matthias Saou wrote :
> > try $name, that should do it, came in somewhere mid 2.6.x series
>
> Yup, that was it!
>
> Section "$name can now be used to set default values in defined
> resource types" in the release notes :
> https://projects.puppetlabs.com/
R.I.Pienaar wrote :
> > Maybe I'm going nuts, but I could have sworn reading about a trick or
> > new feature which allowed to have a definition parameter get a
> > default value of the definition's own $title (instead of having the hack
> > around the limitation with selectors and intermediate va
Hi,
Maybe I'm going nuts, but I could have sworn reading about a trick or
new feature which allowed to have a definition parameter get a default
value of the definition's own $title (instead of having the hack around
the limitation with selectors and intermediate variables later on).
Basically wh
Andreas Kuntzagk wrote:
> I want some config depending on memorysize.
>
> What I tried was
>if ($memorysize >= 256 * 1024*1024) {
> ...
>}
>
> But this fails because $memorysize is a string (and contains a "G")
> and can't be compared to an int.
>
> Are all facts strings? How do I work
Hi,
I very recently upgraded from puppet 0.25.4 to 2.6.8 (both master and
all nodes), and I'm now facing serious memory usage issues. On the
master, I just added more RAM because I had the resources available
(from 2GB to 4GB). But on some nodes with 256MB of RAM, I now see some
load peaks and ser
> On Tue, Jun 14, 2011 at 6:45 PM, Nick Fagerlund
> wrote:
> > On Jun 14, 8:26 am, Matthias Saou
> >
> > wrote:
> >> Could the documentation section about capitalization be updated to
> >> include a scoped example and make it clear for
> >> everyone
Hi,
The documentation section gives a good example for the "exec" type :
http://docs.puppetlabs.com/guides/language_guide.html#resource-defaults
But for scoped definitions, I had to figure out that each word needs to
be capitalized, not just the first word :
Mymod::mydef { 'foo': } gives Syntax
Hi,
Then I'm guessing the only other solution, which also works with 0.25
and any higher version would be :
$local_scope_fqdn = $::fqdn
Then :
<%= local_scope_fqdn %>
In order to use a local scope variable from within templates.
It's really too bad to not be able to use scoped variables from
wrote:
> Try:
>
> inline_template('<%= scope.lookupvar("::fqdn") %>')
>
> ken.
>
> On Sun, Jun 12, 2011 at 3:20 PM, Matthias Saou
>
> wrote:
> > Hi,
> >
> > I'm starting to play with the latest puppet 2.7.0rc4,
Hi,
I'm starting to play with the latest puppet 2.7.0rc4, one of my goals
being to try and update all of my puppet related files to be compatible
with it, with no warnings at all.
I've read in detail this page :
http://docs.puppetlabs.com/guides/scope_and_puppet.html
There is no mention of anyth
Dan Bode wrote:
[...]
> > Comments about what I'm doing right and what I'm doing wrong would
> > be very welcome. I really want to know that everything's optimal
> > before investing time in publishing more modules, to not have to
> > later waste time going over all modules again.
> >
>
> The co
Hi,
I've been using puppet for a while now, and over time I've created many
classes and definitions for a whole bunch of different things.
It's all very RHEL4, RHEL5 (and now RHEL6) centric, but since lots of
people use that (or clones like CentOS), I'm sure it could all be reused
by many. The lic
Steve Shipway wrote :
> 1. The syslod has changed to rsyslog [...]
>
> 2. The snmpd has changed the location of its var file [...]
>
> 3. The use of UDEV for all devices [...]
>
> Has anyone else spotted any other changes requiring alteration to the Puppet
> manifests and modules?
There are pr
Spenser Gilliland wrote:
> Yeap, I'm seeing that it doesn't exist yet either. I'm thinking maybe
> I can use exported resources or maybe just an exec that increments a
> counter on the host. Either way it's not very pretty. My use case is
> as follows:
>
> In Debian, Redmine has the option of
Peter Meier wrote :
> > Is this a known problem with 0.25 clients? Has anyone seem these
> > problems too?
>
> There have been some fixes for it in 0.25.1 [1] but I'm still
> encountering such issues as well. Especially as I'm connecting some
> clients over the internet, so I might encounter
Hi,
Since upgrading from 0.24 to 0.25, I've seen repeatedly puppetd clients
just stop when they're unable to reach the puppetmaster for a while.
With the normal logging level, they just seem to stop without writing
anything in their log.
I've seen this with 0.25.1rc1, rc2 and final, which are al
Brice Figureau wrote :
>
> On Tue, 2009-10-27 at 17:25 +0100, Matthias Saou wrote:
> > I've got the following working fine with puppet 0.25.1, which was
> > previously impossible with 0.24 (yepee!) :
> >
> > define dirs ( $dirs = [] ) {
> >
Hi,
I've got the following working fine with puppet 0.25.1, which was
previously impossible with 0.24 (yepee!) :
define dirs ( $dirs = [] ) {
# Parent directory for the ones below
file { "/nfs/skel/${title}": ensure => directory }
# Make sure the File namevar is uniqu
Todd Zullinger wrote :
> Matthias Saou wrote:
> > When I update my rpm packages on the master (RHEL 5.4), when the
> > "service puppetmaster stop" command is run, the
> > /usr/sbin/puppetmasterd file is removed. Ouch!
>
> Ouch indeed. This is my fault. :(
&
James Turnbull wrote :
> Puppet 0.25.1 - code name "zoot" - is now available. The 0.25.1 release
> is a maintenance release in the 0.25.x branch.
I'm seeing something really weird with this release. I wasn't seeing it
with either rc1 or rc2.
When I update my rpm packages on the master (RHEL 5.
Douglas Garstang wrote:
[...]
> It seems to me like there might be a bug here. Puppet appears to think
> the syslog-ng and syslog service are the same thing and stops
> syslog-ng instead of syslog.
> If I change the name of the syslog service, like this...
>
> service { "XXXsyslog":
> e
Pete Emerson wrote:
> We've got over 150 hosts hitting the one puppetmaster, and based on
> what I've seen via searching it seems like we're hitting into
> scalability issues with Webrick, and the recommendation is to switch
> to Mongrel or Passenger. Looks to me like Passenger is where the focu
william Famy wrote:
> I prefer runing class on my client if thereis a file exemple if the
> file /etc/mypuppet/condition is present execute the condition class.
If you want to do this, you'll likely have to create a simple facter
fact for your clients so that the puppetmaster receives "true" if
Todd Zullinger wrote :
> James Turnbull wrote:
> > Puppet 0.25.1 Release Candidate 1 is now available for testing.
>
> Packages for Fedora 10/11/rawhide and EL 4/5 are available at:
>
> http://tmz.fedorapeople.org/repo/puppet/
>
> Please report any packaging or repository bugs to me and no
Paul Nasrat wrote :
> > My problem here is that facter doesn't seem to be able to return
> > anything other than strings. Is that the case? Because it would be
> > really easier for me to have it return an array of the members :
>
> Correct currently facter is essentially a flat key value map. T
Hi,
As often with puppet, I'm trying to do something I think is "simple",
yet I'm hitting a brick wall.
What I want to do is have a client send back a list of members of a
system group so that the server can create some template-based
configuration files for a specific daemon, one per member.
M
84 matches
Mail list logo
