On 06/08/2012 09:06 AM, jcbollinger wrote:
On Jun 7, 8:29 am, Jistan Idiotjistanid...@gmail.com wrote:
So I discovered yesterday that if puppet changes the gid of a group it
doesn't go through the file system and update them with the new correct gid
(at least this was the case on RHEL5
On 03/12/2012 10:16 AM, Peter Bukowinski wrote:
Your master's version should always be equal to or greater than the
latest client version you're using. You can count on a newer master
working with older clients, but don't count on newer clients working
with older masters.
So, no problems
On 03/12/2012 08:22 PM, MF wrote:
Hello,
I am setting up a new master server and migrating my configuration
from my current master server. I am also upgrading from 2.6 to 2.7.
I have the new server built and integrated with Apache, Passenger, and
Dashboard just like on my current server. I
Can you make your symlinks look like this?
libsomething.so --- libsomething.so.0
libsomething.so.0 --- libsomething.so.0.vv.rr
the base links can change all they want, in that case.
--vagn
On 12/30/2011 11:32 AM, Len Rugen wrote:
We have a case where we've been requested to create an extra
On 08/12/2011 09:32 AM, Chad wrote:
Any ideas?
For starters, see what happens when you run it in an
exec { build_iptables status: logoutput = true, }
--
vagn
--
You received this message because you are subscribed to the Google Groups Puppet
Users group.
To post to this group, send
On 08/11/2011 04:42 AM, Tom De Vylder wrote:
# cat include/init.php
?php
$dbhost = '%= app_dbhost %';
$dbname = '%= app_dbname %';
$dbuser = '%= app_dbuser %';
$dbpass = '%= app_dbpass %';
?
Surely you mean (note the dash (-) ):
?php
$dbhost = '%= app_dbhost -%';
On 08/10/2011 02:40 AM, ki_chi_saga wrote:
exec { 'a': ... } -
exec { 'b': ... } -
exec { 'c': ... }
can I really be sure that puppet evaluates this in the sequence of
writing?
See the little arrows?
-
That is one way to declare sequence. There are others.
See the
On 08/08/2011 05:55 PM, David Kavanagh wrote:
eucalyptus/modules/ntp/manifests/init.pp (defines class eucalyptus::ntp)
try putting
whatever/eucalyptus/modules
on your module path.
--
vagn
--
You received this message because you are subscribed to the Google Groups Puppet
Users
On 08/09/2011 01:19 PM, ki_chi_saga wrote:
To me the Exec[shareall] will be run last? But does it?
Rgds,
Mat
if you care about order you should make it explicit:
exec { 'a': ... } -
exec { 'b': ... } -
exec { 'c': ... }
or
exec { 'a':
before = Exec[ 'b' ],
...
}
Whenever we want to add additional parameters to certain classes,
unless we update that class and the YAML that drives it atomically
everywhere, we end up in a situation where the YAML does not match the
parameterised class and causes errors.
Warning: brainstorm ahead.
As long as you can
On 08/08/2011 02:52 PM, Douglas Garstang wrote:
file {
'/opt/sugarsync/tomcat/tomcat-home/current':
ensure = inline_template(tomcat-%=
$config['tomcat_version_server'] %);
}
Try without the dollar sign:
file {
On 08/08/2011 04:26 PM, Douglas Garstang wrote:
But, I would like to replace:
file {
${tomcat_home_dir}/current:
ensure = inline_template(tomcat-%=
config['tomcat_version_server'] %);
}
with:
file {
$config['tomcat_home_dir']/current:
On 08/05/2011 04:25 AM, Thomas Rasmussen wrote:
Is there any way of overriding a previously include of a class?
Yes. See
http://docs.puppetlabs.com/guides/troubleshooting.html#class-inheritance-and-variable-scope
Something like (warning: untested):
Class[Sudoers::Config] { env =
On 08/05/2011 10:21 AM, R.I.Pienaar wrote:
But also now I think $foo should be considered a top level variable in 2.7
when its not, making node variables more or less useless.
I like it that node variables are local and private.
node default {
$foo = $backupdir/${hostname}
On 08/05/2011 01:18 PM, jcbollinger wrote:
On Aug 5, 9:38 am, vagn scottvagnsc...@gmail.com wrote:
I think one could make a case for getting rid of global variables entirely.
Sort of. There are alternatives that could fill the same role, such
as class variables of a class that
What version of puppet are you on?
On 08/04/2011 03:30 AM, Jean-Christophe Lacaze / Bintz wrote:
It's a fact generated automatically when something is mounted and
replace / with _ to generate the fact variable.
In my exemple, I have a mountpoint on /var/lib/xxx-yyy/zzz that points
to /dev/drbd0
On 08/04/2011 06:18 AM, Jean-Christophe Lacaze / Bintz wrote:
Im on a 2.6.1
from
http://projects.puppetlabs.com/projects/puppet/wiki/Allowed_Characters_in_Identifier_Names
Variables: ΒΆ
On 08/04/2011 09:14 AM, Jean-Christophe Lacaze / Bintz wrote:
Isn't there a way to protect the hyphen
Not that I know of. Have you considered
getting rid of the hyphen? Why does your
mount have to be xxx-yyy? can't it be xxx_yyy?
Anyway, I'm out of ideas on this. Good luck.
--
vagn
--
On 08/03/2011 05:49 AM, Jean-Christophe Lacaze / Bintz wrote:
$mountpoint_path_my-path being the facter variable
${mountpoint_path_my-path}
should work on 2.7.2rc2. Earlier versions
of puppet don't let you have '-' in a variable
name. Not sure if you are allowed to have
a '-' in the fact
On 08/03/2011 12:50 PM, Craig White wrote:
an erb template for the clients' puppet.conf which has environment =%=
environment %
the environment is set on the client, so you've implemented a no-op.
--
vagn
--
You received this message because you are subscribed to the Google Groups Puppet
prep for 2.8: default values and overrides
I was very happy, some months ago,
when I worked out how to separate
configuration from code in puppet.
The scheme even allowed for specialization.
The style was something like this:
cat 04_default.pp
#! /usr/bin/puppet apply
node default {
On 08/02/2011 03:52 PM, Jfro wrote:
Can someone walk me through the steps of sending out a password to all
my computers for user student.
I do this a lot:
yes 'PASSWORD' | passwd username
Then you can extract the password from /etc/shadow
also see here:
prep for 2.8: scopes and variables: nodes are not classes
I have been preparing for 2.8 by
experimenting with 2.7.X and seeing
what it takes to get rid of the
warnings. I would like to share my
results in a number of upcoming posts.
Begin with the simple question of
hard-wiring scopes. the
On 08/01/2011 06:40 AM, David wrote:
The scenario is, every server in our estate of Debian machines
requires some monitoring software which is not provided as a Debian
package.
You should make your own debian package(s) for the monitoring software
instead of installing from source on every
On 08/01/2011 01:41 PM, treydock wrote:
#includedir /etc/sudoers.d
Maybe without the quotation marks?
--
vagn
--
You received this message because you are subscribed to the Google Groups Puppet
Users group.
To post to this group, send email to puppet-users@googlegroups.com.
To
On 08/01/2011 05:04 PM, Jared Curtis wrote:
* Central Dashboard instance
* Central CA
Would it be a problem if your central
CA and dashboard became unavailable?
-v
--
You received this message because you are subscribed to the Google Groups Puppet
Users group.
To post to this
On 07/29/2011 08:57 PM, newguy wrote:
I want to add the ip address of client B in a module M of client A.
Lets assume that I don't know their ip addresses before hand.
Well, what do you know before hand?
Do you know the name of client B? Why can't you use that, directly?
If you know
Funny you should mention that. I have been playing around with
separating the
puppetmaster config from the agent config. See here:
http://agawamtech.com/blog/?p=383
You could probably do something similar, and it would allow you to
switch the agent from one master to another.
--
vagn
In 2.7.2rc2 the '-' is now
parsed as part of the variable name.
Is this a deliberate change?
#! /usr/bin/puppet apply
$hello = hello
$world = world
$hello- = SURPRISE
notice($hello-$world)
notice(${hello}-$world)
# result:
# # ./surprise1.pp
# notice: Scope(Class[main]):
On 07/26/2011 10:32 AM, Nigel Kersten wrote:
Allowing a trailing hyphen doesn't seem like the most optimal solution
though
That doesn't matter so much. But, expanding the character set for
variables without even mentioning it in the release notes has caused me
some pain.
You really
On 07/26/2011 10:54 AM, Nigel Kersten wrote:
I guess it's just a case of being a defensive shell programmer, but
I've always considered curlies-with-double-quotes to be essential with
all interpolated strings.
A better way to handle it would be to make curlies mandatory, and then
change
On 07/25/2011 08:47 AM, jcbollinger wrote:
Puppet should not, and typically doesn't, apply resources that require
a failed resource. The content of the file managed by
File[Make_Sudo_File_Live] is not directly relevant.
I wonder what would happen if he spelled his
dependency chain like
On 07/24/2011 01:40 AM, John Martin wrote:
You must add
a second grep to ignore it.
the idiom is to use a trivial character class:
unless = /bin/ps -aux | /usr/bin/grep '/[d]ata/service',
--
You received this message because you are subscribed to the Google Groups Puppet
Users group.
Node scope is not global scope. But, more interesting:
On 07/23/2011 04:18 PM, rvlinden wrote:
I really love the way I can inherit from nodes and define new or
overrule existing variables.
I have been experimenting with puppet 2.7.1.
It turns out that parameterized classes can inherit
On 07/21/2011 10:57 AM, John Hawkes-Reed wrote:
It seems to me that mixing code and data like that might be a bad idea, and was
wondering if anyone else had a better one.
It would be cool if puppet knew about svn, git, and so on. A file could
have a
URI and revision, when the revision
On 07/22/2011 09:08 AM, jcbollinger wrote:
Well yes, but that's not relevant to the OP's problem. I had in fact
supposed that he omitted all that so as to provide a smaller failure
case.
The OP's problem is that he is not including the header
fragment in Assemble_Sudo_Fragments. It is easy
On 07/21/2011 03:14 AM, Hugo Deprez wrote:
Is there any mechanism available to auto delete old files ?
Or should I configured my own script to clean up old reports ?
I'm doing this:
class puppet::clean_reports {
cron { puppet clean reports:
command = 'cd
On 07/21/2011 08:08 AM, vagn scott wrote:
On 07/21/2011 03:14 AM, Hugo Deprez wrote:
Is there any mechanism available to auto delete old files ?
Or should I configured my own script to clean up old reports ?
I'm doing this:
class puppet::clean_reports {
cron { puppet clean reports
On 07/21/2011 08:52 AM, vagn scott wrote:
Thinking about this some more, if you need the reports for dashboard
to work,
then any cleanup script should leave the latest report from any server,
so that, even if the server has not checked in for a while it won't
disappear.
Thanks
On 07/20/2011 09:37 PM, Jon Jaroker wrote:
Hello, I was wondering if anyone can spot the mistake I am making in
the file-fragments pattern below.
You probably want to drop ALL your fragments into the
fragments directory, including the head fragment.
Use naming conventions to establish
On 07/20/2011 04:43 AM, Matthias Saou wrote:
* The definition is in a generic module for which it does not make
sense to default the parameter to this calculated value.
Instead of passing in the port number you could pass
a template for its creation
node 'myvmhost' {
createvm { [
On 07/20/2011 04:43 AM, Matthias Saou wrote:
node 'myvmhost' {
createvm { [ 'vm01', 'vm02', 'vm03', 'vm04', 'vm05' ]:
# VNC Port 59XY for vmXY
vncport = regsubst($name, 'vm', '59'),
}
}
that looks like an interesting module.
are you planning to publish it?
Or, could I get
On 07/18/2011 06:45 PM, Nick Fagerlund wrote:
You're not the only one -- I've never gotten a puppet apply shebang
line to work. I investigated around the office, and I think what I
found was that everyone remembered it having worked at some point in
the unspecified past, but no one could specify
On 07/19/2011 09:50 AM, rvlinden wrote:
Can somebody explain me what I am doing wrong here ? or what I have
missed ?
WHy does the apache params 'see' the variable frmo the node config and
the users params not ?
bad news: your apache class relies on dynamically scoped variables.
When dyn.
On 07/18/2011 11:27 AM, zu...@puzzle.ch wrote:
I saw that some use the shebang #!/usr/bin/puppet apply to start such
manifests directly as a script. This does not seam to work for me. All i
get is:
If your script name is foo, and you have
#!/usr/bin/puppet apply
as the first line, then you
On 07/17/2011 12:45 PM, S Ahmed wrote:
So I ran through a server setup on ec2, and have a text file of all
the commands I used to get the server to where I wanted it.
step 1: turn your list of commands into an idempotent script
#! /bin/sh
On 07/17/2011 03:05 PM, S Ahmed wrote:
The last bit of what you said I don't understand:
But, the biggest improvement would be to make your own package and
install that
instead of fiddling around with this low level stuff in the manifests.
My apologies. I should remember to use standard
Is suppose I should mention that, if you have an idempotent script like
the one mentioned,
you can use it directly. Again, not tested, so may bugs.
class ruby_from_src {
file { /root/ruby_from_src.sh:
content = template(ruby_from_src/ruby_from_src.sh),
On 07/12/2011 06:19 AM, Al @ Lab42 wrote:
command = mkdir -p $destination_dir ; cd
$destination_dir ; $extract_command $work_dir/$source_filename,
Nice. But I would suggest changing ';' to ''. That way, if the
mkdir or cd fail you don't end up
trying to extract the archive in
On 07/12/2011 01:05 AM, Patrick wrote:
Keep in mind that, this being unix, there's no drives so if you will
probably be checking the amount of space on root if you don't think about it.
There most certainly are drives in unix. Not drive letters, though.
The drives are normally denoted
On 07/11/2011 10:37 PM, Victor Hooi wrote:
As long as we have more the x bytes available when we install
things, we know we'll be fine.
Here is a proof of concept.
With a little tweaking you should be able to test for
an absolute amount of free space on any mounted volume.
Hope that helps,
On 07/07/2011 12:43 PM, romuald FREBAULT wrote:
has anyone an idea of how i could have managed files removed on
destnation when removed from source without using the entierely purge
function?
What you want is a /etc/cron.puppet.d and a way to link files there
into /etc/cron.d (cron supports
On 07/08/2011 04:10 PM, Chip Schweiss wrote:
How to access a variable set in the node declaration from a class that
executes in the post stage?
The answer is the same whether you use stages or not.
Put your variables in a class.Below was tested on
puppet 2.7.1. Per node stuff (eg
The catalog run time that is reported: does that include the facter run
time?
If not, how should I get that? What kind of performance statistics does
puppet/facter keep, if any, and how do I access them? Is there a FM I
should R?
--
Thanks,
vagn
--
You received this message because you
That sounds suspiciously like procedural thinking about a declarative
language.
You should understand that in puppet everything is 'set' before anything
is 'done'.
Stages just help to order the 'get it done' part.
So the answer to your question about a method to access a variable
is:
On 07/05/2011 01:57 AM, Peter Meier wrote:
Can also be:
Yes, I know. Now expand this one, and maybe you will get my point.
-
Thinking about this some more, I like block better,
with order optional:
class c {
block { x:# any order
package {
On 07/05/2011 04:35 AM, Ken Barber wrote:
Whats wrong with using chained resources?
It doesn't scale.
Try expanding this (it is a slightly improved
proposal with block{} instead of order{}):
class c {
block { x:# any order
package { ... }
file
On 07/05/2011 05:35 AM, Luke Bigum wrote:
avoiding the use of a class for ...
conceptual reasons.
Exactly right. Thank you for that phrase. -v
--
You received this message because you are subscribed to the Google Groups Puppet
Users group.
To post to this group, send email to
On 07/05/2011 05:51 AM, Ken Barber wrote:
Then you can also use these anonymous blocks/closures with defaults as well:
class c {
{
Service { hasstatus = true }
service { a: }
}-
Yes, they should have their own scope, and pretty normal
puppet semantics/conventions. Maybe
On 07/05/2011 07:15 AM, Chris Phillips wrote:
debug: Service[puppet](provider=redhat): Executing '/sbin/service puppet
stop'
notice: Caught TERM; calling stop
[root@gibsvlin-erp2dst ~]# /etc/init.d/puppet status
puppetd is stopped
That's just weird looking.
So puppet is running, and
On 07/05/2011 09:24 AM, Martin Alfke wrote:
On 07/05/2011 03:19 PM, Brian Gallew wrote:
I was seeing the behavior on my Solaris boxes when running Puppet under
SMF. The issue, in my case, was that I was trying to work around an SMF
bug. My workaround was to svcadm disable
On 07/04/2011 02:49 PM, Michael Halligan wrote:
Forgive my cluelessness, but why is it so difficult for Puppet to at
least tell me what file this came from
OK, I'll take a crack at an answer.
---
typical compiler versus parser problem.
If the parser saw this, it could stop
Sometimes things just have to happen in sequence.
It is the simplest of relations, but puppet really
has no convenient, non-fiddly way to express it.
So, how about
class x {
order { z:
file { a:
...
}
exec {
On 07/04/2011 11:42 PM, Scott Smith wrote:
I take it you're not familiar with the `requires' parameter.
I'm familiar with
requires
subscribe
notifies
-
-
before
which are great for specifying relations between non-adjacent things.
But they are annoying to use for
On 07/04/2011 11:51 PM, Scott Smith wrote:
Ignoring the fact that noise is very subjective... it doesn't. In
Puppet, position in a file has no bearing on order. It's not an
imperative language.
I understand. What I am saying is that another way to express this
file { a:
On 07/04/2011 11:56 PM, vagn scott wrote:
On 07/04/2011 11:51 PM, Scott Smith wrote:
Ignoring the fact that noise is very subjective... it doesn't. In
Puppet, position in a file has no bearing on order. It's not an
imperative language.
Oops, that should be:
file
Is the client healthy?
File system not full?
No errors in log files, like a disk might be failing?
Load average reasonable?
Any network changes?
Do you have a switch plugged into itself?
what does tcpdump show? Normal activity?
--
vagn
On 07/05/2011 12:11 AM, sanjiv.singh wrote:
some times
of a
resource?
On Mon, Jul 4, 2011 at 8:03 PM, vagn scott vagnsc...@gmail.com
mailto:vagnsc...@gmail.com wrote:
$foo = inline_template( ... )
$baz = inline_template( ... )
File[ $foo$baz ]
--
You received this message because you are subscribed
Thinking about this some more, I like block better,
with order optional:
class c {
block { x:# any order
package { ... }
file { aaa: }
exec { bbb: }
file { ccc: }
include foo
}
block { y:
On 07/05/2011 01:08 AM, Scott Smith wrote:
Whatever problem you are trying to solve is most likely best handled
outside of Puppet.
The problem I'm trying to solve is one of
expressiveness in the language. I'm not
adding any functionality that is not already
in puppet. I'm just suggesting
On 06/26/2011 02:01 PM, ssk1287 wrote:
Consider the following scenario.
a folder FOLDER 1 which is owned by the root as it should be
accessible to all users of the system.
That's your mistake, right there. If you want a folder
accessible to all users then make it group writable for
some
On 07/02/2011 03:20 PM, Eric Sorenson wrote:
This is usually because somewhere you have a File resource whose name
is a variable, and the variable's undefined.
And a common reason for that to happen is that you've
mistyped the variable name. Your design could be just fine.
This is a good
Maybe you've got network issues?
1 minute sounds like two DNS lookups timing out.
--
vagn
On 06/30/2011 01:42 PM, Haitao Jiang wrote:
Thanks a lot for both answers! Very helpful.
However, I was a little surprised that how slow Puppet was when I run
your example. Was it Puppet just slow or was
On 06/17/2011 10:52 AM, Nigel Kersten wrote:
It's not entirely unreasonable to expect that if module 'foo' is found
in one modulepath component, that no other module named 'foo' should
come into play, but I think we have some edge cases here. Please feel
free to file a feature request around
On 07/01/2011 10:01 AM, jcbollinger wrote:
Parameterized classes address problem (1) by formalizing class
parameterization, and they address problem (2) by making it illegal to
include a parameterized class more than once (even with the same
parameters). One of my main objections to
Here's one way. It's ugly, but it works.
--
vagn
#! /usr/bin/puppet apply
# Hello everybody
#
# I have a set of nested classes and the deepest class looks after a
# file resource. I want all my nodes to be of the parent class type
# except one of them needs a special copy of the same file
does /var/run/puppet exist?
-v
On 06/21/2011 12:30 PM, Craig White wrote:
I am using ruby-enterprise for ruby packages and gem for puppet/facter
installation and I 'borrowed' the ubuntu sysv init scripts and they mostly work
but for some reason, it won't stop the service as written but it
I'm doing something similar with class inheritance,
rather than node inheritance.
In http://docs.puppetlabs.com/guides/scope_and_puppet.html it says:
In effect, all variables will be either strictly local or strictly
global. The one exception will be derived classes, which
you also have to have a
include foo1
somewhere.
Note that your require is a dependency,
not an instantiation. require just says
that foo1 has to be installed first,
before libreoffice package can be installed.
--
vagn
class libreoffice {
case $hostname {
foo: {
On 06/29/2011 09:56 PM, treydock wrote:
% apparray.each do |key,value| -%
Key:%= key %
Path:%= value['path'] %
Command:%= value['command'] %
% end -%
inline_template() can be used as a here document.
that plus a puppet shebang line makes testing
and presenting examples really easy.
Put the
In my reading I just ran across this:
[Puppet - Bug #7273] Modifying puppet.conf causes 'reparsing
config'
and TERM signal results in shutdown of daemon
short version: puppet agent can commit suicide, but it can't raise
itself from the dead.
However, init is immortal, and
Have you looked at clusterssh?
--
vagn
On 06/21/2011 12:01 PM, Andreas Kuntzagk wrote:
CHEBRIAN wrote:
Hi,
how to install puppet client thru puppet master .
My requirement is to install puppet client nearly 400 client servers.
its difficult to install puppet client in all the servers by
On 06/18/2011 10:31 AM, Markus Falb wrote:
On 18.6.2011 05:30, vagn scott wrote:
or you can do what I do and pipe that archive into tar:
exec { true $title wget -O - $upstream | tar xzf - --xform
's,wordpress/,,':
cwd = $top,
creates
On 06/17/2011 09:14 AM, lucas.brig...@ymail.com wrote:
Hi!
I'm doing a module for the puppet for WordPress will implement.
I'm having problems with this line:
exec { tar -vzxf ${downdir}wordpress.tar.gz:
path =
[/usr/local/bin,/opt/local/bin,/usr/bin,/usr/sbin],
Let's see: Then I could do
Stage { ensure = enforcement, }
at the top and get the behavior I want:
prerequisites are satisfied before moving on.
But, what's the use case for relationship?
Why would I want that?
consider three stages in various
combinations of enforcement (e) and relationship
On 06/11/2011 01:06 PM, Nan Liu wrote:
I'm not sure if I captured the intention well, essentially still want
a way to coarsely organize classes, however without adding any
dependency requirement. So deploy user accounts/customization after
application deployment, however still proceed with
Puppet already has stage[ Main ] which is the only
stage it needs to define. All other stages
can be defined relative to main and each other, and should
be a matter of convention. So I think it would be more
fruitful to talk about the purpose of stages, along with
their proposed names.
For
On 06/10/2011 08:55 AM, jcbollinger wrote:
Maybe what is needed is an easier way to perform two or more
successive puppet runs in different environments.
It would be nice to have a puppet restart command.
Tell puppet that it should reinitialize and try again.
Then you could, as you say,
A while back I wrote down all the puppet patterns
I could think of, and this was one of them.
I named it Cradle To Grave, but probably that's not
appropriate. However, I was only focusing on
puppet runs at the time, so that name popped into my head.
It is an instance of the more general pattern
|Does this help?
dpkg -L PACKAGENAME
|
On 06/08/2011 01:44 AM, Robin Lee Powell wrote:
(zombie thread rar!)
Where this comes up for me is when I have packages set to latest.
There's not really any way, I don't think, to integrate samhain into
this process (that is, to say I just
What should this do?
$array = [ one, two ]
$greeting = [ hello, howdy ]
@file { host_$array$greeting }
or this?
$array = [ one, two ]
@file { $array_blah }
seems like we would need delimiters
host_${array}_${greeting}
and some form of expansion control.
Should it
It would be convenient if you could download from a debian repository
that has the packages you want. The logical place to find that
is somewhere under the control of the Amanda people,
but they don't seem to have that set up. The debian backports
site doesn't have the version you are
On 04/15/2011 03:05 PM, Rob McBroom wrote:
On Apr 15, 2011, at 2:58 PM, Ramin K wrote:
I use the IP address to generate a 32bit integer for the server_id for
my Mysql configs.
server_id =%= ipaddress.split('.').inject(0) {|total,value| (total
8 ) + value.to_i} %
Would something like
It sounds like you are solving the wrong problem.
Why do your users have root privs?
If they need root for some things, use sudo to give them
only what they need. Or use suid/sgid mechanisms
to allow community access to certain resources.
If you need to lock down specific files look at the
I took another pass at this and it seems to work
the way I expected it to. I'm not sure why it didn't work before,
and I don't have time right now to look into it.
Anyway, thanks to those who replied. If I find anything
interesting I will post it.
--
thanks,
vagn
On 04/29/2011 10:53 AM,
Please look at
ldconfig -N
it will update the links for you.
You could try something like (warning untested)
class foo {
package { liblcgdm:
notify = Exec[ ldconfig::update ],
}
include ldconfig::update
}
class ldconfig::update {
exec {
On 04/29/2011 10:53 AM, jcbollinger wrote:
On the conceptual level, what does it or *should* it mean that a node
has Class['proxy']? To you it seems to mean that the node positively
uses an HTTP proxy, at least for Apt. But it could instead mean that
the node's proxy configuration is
On 04/29/2011 12:29 PM, R.I.Pienaar wrote:
Sounds like you want resource chaining[1]
careful though this has an unfortunate side effect that it
will also realize all virtual packages with the provider apt
onto a node.
point is you *can* tweak dependencies as a result of including
a
I didn't run this (simplified) example so there might be errors, but
anyway -- say we have:
---
stage { aptstuff: }
Stage[aptstuff] - Stage[main]
class proxy($url) {
file { /etc/apt/apt.conf.d/02proxy:
content =
Oh, great timing!
I'm working on pretty much the same question. I have used FAI and
Opsware, and homebrewed scripts to do provisioning in the past. I have
never been all that happy with any of them, and am taking another look
at the problem to try to get to as simple and portable a solution
100 matches
Mail list logo