Very nice! And since I thought I had found a bug (I ultimately learned I
hadn't), I did some tracing and now know more about vim highlighting, which is
a double bonus!
On Jul 21, 2012, at 5:40 AM, no-re...@cfengine.com wrote:
> Forum: CFEngine Help
> Subject: vim syntax highlighting
> Author:
Jason-
Stay tuned for an announcement of a USENIX "Configuration Management Summit -
Virtualization, the cloud, and scale" being help in Boston in June. Your
opinions and ideas would be extremely welcome there!
-Dan Klein
On Apr 13, 2012, at 7:48 PM, no-re...@cfengine.com wrote:
> Forum: CFE
It is pretty hard to determine what you did wrong when you haven't shared what
you did...
Humorously:
before the job:
Bank Account: $192.37
after the job:
Bank Account: $193.42
expected:
Bank Account: $3,987,245.22
Can you discover what did I do wrong?
On Mar 5, 2012, at 4:37 PM, Nick Anders
Jonathan-
Looking naively at FnCallNow() in evalfunction.c, I believe you are correct in
your suspicion that cf-agent uses the start time as "now". Now, I say
"naively" because I am not that familiar with the Cfengine core code, and I
have a not-quite up-to-date version. For a long-running ag
Simon-
As stated, the class you're setting is global in scope. However, the test for
the existence of the class might not be redundant, depending on what exactly
you _really_ want to do.
My suggestion is to use classes whose names are also base on the parameter
variable, i.e., if_repaired("ec
This is a "classic" mistake (that is, everyone makes it) so don't feel badly :-)
The insert_lines promise really says "I promise that this line will be present
in the file, and if it isn't, then I will insert it". So there are a few
things going on here:
1) You said "insert a '{' if it is miss
Neil-
You should look at the tests/ directory in the source tree - there are plenty
of examples of one bundle testing another :-)
What I hear you saying is that you want repeatable randomness - is that
correct? If so, store your results in a file (possibly generated by a script
in a different
You probably want to look at "promise_repaired" at
http://cfengine.com/manuals/cf3-reference.html#classes-in-_002a - but note that
this applies to the WHOLE promise, so you should be careful where you set the
class (that is, in the main promise, in a subcomponent, etc). But FWIW, I use
this fe
Both of your ideas should work, but CFEngine is rather picky about quotes...
Try:
"class_ab" or => {"class_a", "class_b"}
"class_ab" expression => "class_a|class_b";
-Dan
On Sep 27, 2011, at 12:13 PM, D'Ausilio, John wrote:
> I define some classes in a common bundle represe
Nice way to collect all the documentation!
On Sep 22, 2011, at 3:40 AM, Aleksey Tsalolikhin wrote:
> There is a lot to know about CFEngine, which can make it hard for
> people new to the subject. The purpose of this guide is to lay out
> the resources available to CFEngine students and to orient
Before you get started on this, I think you should ask yourself this:
1) Do I want to maintain all of crontab (regardless of which machine gets what)
with CFEngine, or
2) Do I want to maintain a PART of crontab with CFEngine (and handle machine
differences manually on each machine)
Personally,
Neil is correct, of course, but if you start thinking in those terms, then
you're still thinking like Cfengine is a programming language, and not a
declarative language. Why do you need a particular element of a list? The
idea behind lists is that you address all of the elements, either serial
On Aug 13, 2011, at 9:38 AM, Diego Zamboni wrote:
> Hi Tom,
>
>> 1) Are you aware of a wrapper, unique shell or simiar tool that could
>> provide root level access at a read only level?
>
> What comes to mind is to put the read-only functionality you want in a
> specific program, and then giv
Nice stuff!
On Aug 10, 2011, at 5:52 AM, no-re...@cfengine.com wrote:
> Forum: Announcements
> Subject: STIG compliance policy example
> Author: CFEngine
> Link to topic: https://cfengine.com/forum/read.php?1,23071,23071#msg-23071
>
> CFEngine releases policy example for STIG compliance: http://
Aleksey is correct here - a string, which may contain variables to be expanded,
is interpreted as a class, and the promise is evaluated if the resulting
classname is true.
So if the variable "foo" contains the 3 letters "baz", then ifvarclass =>
"foo$(foo)" will allow the promise to be execut
Really? Huh... that's weird... I'd file a bug report with an example, if I
were you...
-Dan
On Jul 15, 2011, at 2:56 AM, no-re...@cfengine.com wrote:
> Forum: CFEngine Help
> Subject: Re: ${this.promiser} in comments
> Author: svenXY
> Link to topic: https://cfengine.com/forum/read.php?3,22798
Sven-
This was reported as a bug on 2010/02/26 and reported as partially resolved
2010/03/19 in https://cfengine.com/bugtracker/view.php?id=58 - here is the
resolution notice: "This can only be partially fixed. The reason is that the
precise knowledge of this.promiser comes too late to use in t
Yep that looks like a bug (so please file a report :-) However, I'd point out
that copying recursively to '/' can be dangerous... (but should be allowed
anyway)
On Jul 7, 2011, at 7:32 AM, no-re...@cfengine.com wrote:
> Forum: CFEngine Help
> Subject: files: copy to "/" does not seem to work
>
I'd do it as a two stage process:
1) Delete any line that has "(#\s*)?O QueueLA" . If you really want to be
fancy, I think you can say say "(#\s*)?O QueueLA=(?!100)" to only delete if the
value is not 100).
2) Insert "O QueueLA=100"
Remember, cfengine does not edit "the file", it edits a temp
One way is to specify a location attribute, or use the insert_type =>
"preserve_block" attribute
-Dan
On Jun 29, 2011, at 10:13 AM, no-re...@cfengine.com wrote:
> Forum: Cfengine Help
> Subject: insert_lines twice problem
> Author: schattenfell
> Link to topic: https://cfengine.com/forum/read.p
You're using the "comment" regex to match lines that DON'T look like
something... Try this:
"dim_array" int =>
readstringarrayidx("array","datafile","^(?!VM;[^;]*;name1;).+",";",10,1);
The regex says "anything, as long as you don't seeVM;stuff;name1;, is a
comment". I think it'll work...
Aleksey/Seva-
So you want to recreate "make" in CFEngine? :-)
ifvarclass() says "if the named class exists, do something"
fileexists() says "if the named file exists, make the class expression 'true'"
and that is used to create a named class.
The two are incompatible (because the latter is onl
I agree with Mark, and the error message is pretty much spot-on.
-Dan
On Jun 18, 2011, at 1:56 AM, Mark Burgess wrote:
>
> This is not a bug, I think. The problem is you wrote
>
> "atftpd_conf" string => { "5.2" };
>
> You need either
>
> "atftpd_conf" string => "5.2" ;
>
> or
>
> "atftpd
I agree with Mark, and the error message is pretty much spot-on.
-Dan
On Jun 18, 2011, at 1:56 AM, Mark Burgess wrote:
>
> This is not a bug, I think. The problem is you wrote
>
> "atftpd_conf" string => { "5.2" };
>
> You need either
>
> "atftpd_conf" string => "5.2" ;
>
> or
>
> "atftpd
Also note that if you really do want to indent the string, there is a
whitespace_policy that might help...
-Dan
On Jun 18, 2011, at 4:59 AM, Jonathan Clarke wrote:
> Mark, I think sauer is referring to this typo in the Solutions Guide, at
> http://cfengine.com/manuals/cf3-solutions.html#Set-up-
Well, you could always do this instead:
body file_select modified_over_a_year_ago
{
mtime => irange(ago(1,0,0,0,0,0),now());
file_result => "!mtime"; # modified more than a year ago
}
On May 21, 2011, at 10:16 PM, Aleksey Tsalolikhin wrote:
> The following policy selects files more than a yea
I was just suggesting things to look at. But yes, an index with white space
(or with white space stripped with regextract()) might do the trick...
-Dan
On May 20, 2011, at 4:33 PM, no-re...@cfengine.com wrote:
> Forum: Cfengine Help
> Subject: Re: Cfengine Help: question about editing smb.conf
In general, take a look at whitespace_policy in insert_lines for some ideas...
Also look at set_variable_values in cfengine_stdlib.cf for others
-Dan
On May 20, 2011, at 3:16 PM, no-re...@cfengine.com wrote:
> Forum: Cfengine Help
> Subject: Re: Cfengine Help: question about editing smb.conf
>
Aleksey-
The promise is to run the transformer, nothing more. Most transformers use
$(this.promiser) as their input. For example, you could gzip the file if it
existed, knowing that gzip would remove the plain file once it compressed the
result (so effectively, you promise that if the file ex
There are a couple of slightly different ways of doing this in PCRE:
1) '.*(?i)cfengine.*'
2) '.*(?i:cfengine).*'
The first one turns on the "/i switch for everything after it, so the .* is
also case-ignored
The second one turns on the "/i switch" only for "cfengine".
There's no difference in
One thing to add: processes are first selected by the name in the promise, and
then by process_select (which may be missing, in which case you just get what
was in the promise). All other actions are in the promise.
So typically, you'd do one of two things:
1) Select "foo" in your promise nam
On May 7, 2011, at 11:20 PM, Jerome Baum wrote:
> Also, you can already do "real math" -- call a shell command to do the
> computation and use an appropriate return code, then define the class based
> on that. If the command is pure (as in "pure function") then it fits in with
> promise theory
l it works. So like Miyamoto Musashi says in "The Book of
Five Rings": think hard on this.
-Dan
On May 7, 2011, at 10:54 PM, Jerome Baum wrote:
> On Sun, May 8, 2011 at 04:49, Daniel V. Klein wrote:
> Some business models use a 3-year lifecycle. Amortization and probably other
Some business models use a 3-year lifecycle. Amortization and probably other
stuff. Since you can't do math directly in Cfengine, a "lifecycle index" is a
quick way of getting a value based o the year, for evaluating lifecycles.
Personally, I have no use for it - but other folks do!
-Dan
On
Actually, cf-promises WILL parse the file for syntax errors. While not exactly
the case, classes are "run time" and syntax is "compile time" (except, as you
correctly observer, when classes are used for selective compilation). So you
could do this:
cf-promises -Dmonitor
and get your
A different solution would be to use delete_select and
delete_if_not_match_from_list to eliminate the lines you don't want, then
insert the ones you do. I am not saying this is better, but if you see
different results (like, the problem doesn't "just happen" any more), then
that'd help narrow
It appears you have a slight misunderstanding on indices, etc. Here's how I
would rewrite your code (I didn't test, but you seemed to have a spurious : and
the wrong slist indexing the wrong array (there is no mypasswd_user array, it
is the slist into the mypasswd_array array):
bundle agent pa
The promises look good, and AFAICT you should always get 2 lines, but I'd
suggest the following change for you:
vars:
any::
"dns" string => "192.168.1.20";
machineX::
"search" string => "domain2";
!machineX::
"search" string => "domain1";
This makes it clearer
Ad I stand corrected on the significance of the umask here :-) I
overlooked the fact that the file was created if it didn't exist.
-Dan
On Feb 2, 2011, at 3:57 AM, Eystein Måløy Stenberg wrote:
> I looked into this, because it was taunted so "critical".
> You are exactly right, Seva.
>
Umm... "go+r" does not mean "mode = 044"! It means "add r to g and o" or in
other words, "mode = mode | 044". The umask should have nothing to do with it,
as that is only used for file creation, AFAIK.
As to why they said, "go+r" instead of "644" is anyone's guess :-)
-Dan
On Feb 2, 2011, at
Even better... why not help the developers by checking out the test framework,
and when you find a bug, create a test that demonstrates the bug by failing,
but which will also pass if the bug is gone.
-Dan
On Jan 7, 2011, at 10:56 AM, Mike Svoboda wrote:
> Well, Cfengine is a mature product,
Are you certain that the pattern matches? I see an \{ and no matching \} in
your promise. Likewise, I see spaces in your pattern - are the number and type
of whitespace an exact match? You might want to consider using \s+ or \s*
instead of just a single space.
-Dan
On Jan 7, 2011, at 10:48
You want a commands promise that is a "module". See
http://www.cfengine.org/manuals/cf3-reference.html#module-in-commands. You
can't exactly inherit shell variables, but this is pretty close to what you
said you need, and probably what you really want.
-Dan
On Dec 28, 2010, at 10:05 AM, no-re
Jim-
Great example, and nice clear concise description - I will add a test to the
testing tool to check for this...
-Dan
On Dec 21, 2010, at 3:05 PM, Jim Lawson wrote:
> Hi help-cfengine,
>
> Happy winter/summer solstice (depending on which hemisphere you live in)!
>
> I am trying to create
ich is not a list? Could that avoid the bug
> with "singleton lists"?
>
> Cheers,
>
> Sandra
>
> On 12/17/2010 02:19 PM, Daniel V. Klein wrote:
>> I did not read your config in detail, but you said "only one server in a
>> class" and th
I did not read your config in detail, but you said "only one server in a class"
and then I see lists, so... If you are running 3.1.0(?), you have possibly
stumbled across a bug with singleton lists that is repaired in the versions
after that. Upgrade to the latest version, and you should see th
Duh! Why didn't I think of that?
It was the -f that had me stumped. Good call!
On Dec 9, 2010, at 4:43 AM, no-re...@cfengine.com wrote:
> Forum: Cfengine Help
> Subject: silly cfengine trick #!
> Author: matter
> Link to topic: https://cfengine.com/forum/read.php?3,19692,19692#msg-19692
>
>
All PCRE matches are greedy, unless you make them stingy. So .* is greedy, .*?
is stingy
On Dec 8, 2010, at 12:02 AM, Daniel Dehennin wrote:
> Michael Potter writes:
>
>> I believe the problem is this expression:
>>
>> "^(smtp_host)\s+.*$" replace_with => value("$(match.1) $(smtp)"),
>>
>>
This seems like an awful lot of work to get the long/short hostname. What
about:
classes:
"xxx" expression => regextract("([\w-]+)\..*", "$(sys.fqhost)",
"short_host");
After this, you have the short hostname in $(short_host[1]), and the long one
has always been available as $(sys.fqhost)
The problem that I see with this is that it turns Cfengine into more of a
"push" technology. By defining a class, you're still letting the remote agent
set the policy, but by saying "run this bundle", you're bordering on setting
the policy remotely...
-Dan
On Nov 26, 2010, at 6:29 PM, no-re..
${foo} and $(foo) are equivalent, and @{foo{ and @(foo) are also equivalent.
On Nov 25, 2010, at 2:43 PM, Seva Gluschenko wrote:
> I'm not Cfengine guru by any kind, but it seems like wrong syntax to me. Try
>
> "foo" slist => @(bar);
>
> (parenthesises, not braces).
>
> By all means, Cfengine
It might help everyone to think about regex's this way:
Every repeating pattern (that is, with a *, +, or ?) is "elastic". WHich means
that it will gobble up as much as possible, leaving the remainder of the string
for the rest of the pattern to attempt to match. If the match occurs, fine.
B
Michael, you are right and I feel like a fool. Of course - the regex was wrong!
But to meet you half way... the \s* matches the spaces the first time, and the
[^0] matches the "1". After that, it replaces the 0 with a 1, and then you're
100% right. Doh!
-Dan
On Nov 19, 2010, at 5:50 PM, Mic
First of all, thank you for the self-contained example! That helps a lot.
As far as I can tell, you have found a bug (but also a mistake in your
interpretation). Your pattern wont match in the string "enabled = 00", but not
because you only have a single 0 (it's because the first "0 doesn't mat
Chris-
Well, actually I think Cfengine is doing what it was supposed to, then :-)
Rather than resort to hoop-jumping, why not just get in the habit of executing
your policy immediately after you check it in, to test for missed bugs?
-Dan
On Nov 19, 2010, at 10:17 AM, C R Ritson wrote:
>> I ag
Chris-
I agree with Seva about hard classes. But this makes me ask: why are you so
erratically validating your promises? Why not every 5 minutes, which is the
default? Cfengine is lightweight, and does not load your system. But if you
have "expensive" promises, just change the "if_elapsed"
You could use:
body link_from example
{
when_no_source => "force";
}
But why worry if you're copying the whole tree? Cfengine will converge on the
correct solution, and the ultimate goal is what you seek, and Cfengine will
provide that.
-Dan
On Nov 17, 2010, at 12:41 PM,
There is a bug that has been reported in the latest release involving slists
with a single element. I am sure that a patch is on the way soon...
-Dan
On Nov 16, 2010, at 12:11 PM, Gusto wrote:
> Hi Folks,
>
> I am seeing this behavior on the the 3.1.0 release, where if I define
> in a bundle
Try something like this:
body common control
{
bundlesequence => { "test" };
inputs => { "/var/cfengine/inputs/cfengine_stdlib.cf" };
}
bundle agent test
{
files:
"/tmp/sshd_config"
edit_line => uncomment_lines_containing("PermitRootLogin.*","#"),
classes => if_repaired("restart_ss
No, I was worng (doh!). I am did more research into this, and it is not the
selection issue as I suspected. I apologize for any misleading.
There is something wrong with your range - 15 is not "15 minutes" :-) You
should use "accumulated" on both sides of the irange.
On Nov 12, 2010, at 4:32
It is probably not a bug - it is a language (in)consistency. Well, maybe
that's a bug... let's see what Mark says?
You are promising two contradictory things, and that is the basic problem.
The first promise is that you will set a class "user_procs_out_of_range" if the
number of processes matc
Good catch, Michael - but here's the weird bit - it turns out that the list is
actually there, you just can't iterate over it!
-Dan
body common control {
bundlesequence => { setup, atest };
version=> "0.01";
}
bundle agent setup {
files:
"/tmp/foo" create => "true", edit_lin
as to begin vith an
>> empty slist (now that we can use "cf_null") and then add bundles
>> and inputs for each class the client belongs to. But it's not
>> possible to use the slist as in:
>>
>> "specific" slist => { @(specific) , "other_stuf
On Nov 8, 2010, at 8:02 AM, Marc Baudoin wrote:
> Daniel V. Klein écrit :
>> IMHO, the better strategy is to put everything (common and specific) in the
>> inputs
>
> I'd rather have only what's needed for each type of host instead
> of loading everythin
You could always try -s :-)
On Nov 1, 2010, at 5:21 PM, Jesse Becker wrote:
> On Mon, Nov 01, 2010 at 03:10:27PM -0400, Daniel V. Klein wrote:
>> Try using grep -q, it is much easier than redirecting
>
> But not portable. The -q option is a GNU extension.
>
> [je...@titan
Try using grep -q, it is much easier than redirecting
On Nov 1, 2010, at 3:07 PM, JIM AMANATIDIS, BLOOMBERG/ 731 LEXIN wrote:
>
> Okay, this seems to break with the string being searched is wrapped in
> quotes, regardless of how many "\" surround it to escape it:
>
> classes:
>"Has_ce_nic
It makes sense to do something if a file exists or not (you can do that now),
and likewise if a file is of zero-size or not (not currently possible, but a
non-existent file should also have zero size), so it seems logical to also do
something based on a file's size (i.e., /foo/baz is bigger than
I guess that depends on what you mean by multidimensional arrays... :-) But
this works:
body common control
{
bundlesequence => { "test" };
nova_edition::
host_licenses_paid => "5";
}
bundle agent test
{
vars:
"a" string => "a";
"a[1]" string => "a sub 1";
"a[1][foo]" string => "a
Michael is certainly correct, but the other side of the answer is: why worry
about it? Cfengine is designed to be convergent - if you make a change to your
masterfiles, your inputs will be updated soon enough.
If you're worried about testing your changes, then the better way to do it is
have a
This will work as of 3.1.0, but do it this way instead (auto-iteration is
always through a scalar reference to a list):
bundle agent some_bundle {
vars:
"bundle_list" slist => some_function_generating_slist(...);
methods:
"any" usebundle => "$(bundle_list)";
}
-Dan
On O
Yes, it is a bug which has since been fixed. I'd recommend upgrading to a
newer version of Cfengine (and 3.1.0 will be coming out soon, too!)
-Dan
On Oct 25, 2010, at 2:06 PM, Jean-Noël Rivasseau wrote:
> Hello,
>
> I want to delete a directory and all its content, I am running cfengine 3.0.3
Weird - the email version stripped the stuff in square brackets! There seems
to be a bug in the mail system conversion...
-Dan
On Oct 21, 2010, at 10:56 AM, Seva Gluschenko wrote:
> Daniel, look into forum. They are arrays.
>
> 2010/10/21 Daniel V. Klein :
>> Your variab
Your variable "array1" and "array2" are not arrays - they are scalar strings.
On Oct 21, 2010, at 10:12 AM, no-re...@cfengine.com wrote:
> Forum: Cfengine Help
> Subject: variable expansion inside a single bundle
> Author: bglomm
> Link to topic: https://cfengine.com/forum/read.php?3,18819,18819#
Look at "empty_file_before_editing". You need to know that cfengine doesn't
actually empty the file that you are promising - it makes a copy, does the
edits on the copy, and overwrites the existing file if it needs to. So when
you say "empty_file_before_editing", it starts the edits on a clean
babudro-
It _is_ confusing, but the explanation is simple:
In one edit cycle, Cfengine considers the insertion of each line, one at a
time. If it sees a line already present that it is supposed to insert, it does
not insert it twice. So each of your blank comment lines are the same - and
onl
As far as I know, library.cf is no longer part of the Cfengine distribution.
Th only file that you should have is cfengine_stdlib.cf - if there are any
conflicts, you should just change your private version of library.cf.
-Dan
On Oct 13, 2010, at 7:09 PM, Aleksey Tsalolikhin wrote:
> Hi. Thi
Well first of all, look in cfengine_stdlib.cf - you might want to simply create
a modified version of set_variable_values to do what you want (where you change
the delimiter to "\s+" or something.
But more importantly, you have a fundamental misunderstanding of how edit_lines
works. Here's bas
One simple solution would be to define an array that maps service to process...
What I show below is a quick kludge (if /etc/init.d/bind fails, then try
/etc/init.d/bind9), but you could be neater and make something that uses
isvariable(). Or you could simply define an array element for *every
Jesse-
A very simple solution to your problem is to define the classes in a common
bundle that gets executed before the agent bundle the defines the variables.
It isn't convoluted, but it might be counterintuitive if you're used to a
procedural language (and Cfengine is declarative, not proced
Eduard-
I have been developing a "maintain user accounts" system with Cfengine (so that
you can add users and initialize their dotfiles, do password distribution or
allow users to change their passwords, mark some fields as static and others as
changeable, etc). This will be released very soon
Bernhard-
It would help if I understood better what you are trying to accomplish with the
array... but a few things:
1) Look at section 11.1.2 at
http://www.cfengine.org/manuals/cf3-reference.html#Introduction-to-functions
for building array's from data
2) The error you're getting on this line
Yes, I will make sure it is clearer... maybe not 72pt, though :-)
On Sep 16, 2010, at 4:44 PM, Jesse Becker wrote:
> Is there a way to have this Seva's comment below integrated into the
> docs, preferably in 72pt type? :-)
>
> Subtleties like this are nearly impossible to find in the current
>
What some people have done is to limit the number of bundles in the
bundlesequence to the common stuff, but have one of the bundles call a bunch of
modules:
body common control
{
bundlesequence => {"update", "classes", "security", "host_specific", "tidy",
"backup" };
}
bundle agent host_speci
Nope. What you're forgetting is the DeMorgan Laws.
!A & !B <=> ! (A | B)
!(A & B) <=> !A | !B
-Dan
On Sep 14, 2010, at 2:16 PM, no-re...@cfengine.com wrote:
> Forum: Cfengine Help
> Subject: Logic headache
> Author: neilhwatson
> Link to topic: https://cfengine.com/forum/read.php?3,18232,182
On Sep 13, 2010, at 3:14 PM, no-re...@cfengine.com wrote:
> Forum: Cfengine Help
> Subject: Re: Questions about the reference manual
> Author: mwlarsen
> Link to topic: https://cfengine.com/forum/read.php?3,18107,18209#msg-18209
>
> neilhwatson Wrote:
> --
Actually, you're getting a different error. The syntax error
cf3:/var/lib/cfengine3/inputs/promises.cf:64,25: syntax error, near
token '@'
is the first thing you need to look at. Everything else comes from it.
The problem is that, like it or not, all variables in Cfengine must currently
Nima-
Comments inline, below...
-Dan
On Sep 11, 2010, at 11:03 PM, no-re...@cfengine.com wrote:
> Forum: Cfengine Help
> Subject: Handles vs Classes - Scope Intersection
> Author: nima
> Link to topic: https://cfengine.com/forum/read.php?3,18170,18170#msg-18170
>
> Are we free to do the follow
19 PM, Tod Oace wrote:
> On Sep 1, 2010, at 11:53 AM, Daniel V. Klein wrote:
>
>>> /^($(regexes))$/ # The slist contains strings, but they are
>>> expanded and then the result is interpreted as a regex
>
> I like that. It's very familiar syntax, and solv
Mark-
I'm Cc'ing this to the mailing list so that the rest of user community can
chime in (since I may be barking up the wrong tree). This is in response to
the many bug reports I see which involve regex's (that are in fact, not bugs
but misunderstandings as to what are regex characters). It'
Will-
There is another way to do this (although Mark is certainly correct in what he
says), and that is to do the management of the differences outside of Cfengine:
1) Create two subdirectories: $(masterfiles)/everyone and
$(masterfiles)/special_everyone
2) Populate both those directories with
Correct me if I am wrong, but there does not appear to be a bundle named
"crash" anywhere, right?
-Dan
On Jul 29, 2010, at 3:49 AM, Bas van der Vlies wrote:
> On 28 jul 2010, at 17:28, Bas van der Vlies wrote:
>
>> Hello,
>>
>> cfengine 3.0.5, did not see anything in the Changelog if this is
Heh - just use "set_variable_values" from cfengine_stdlib.cf! Or see how it
is done there - it's rather clever (I wish I thought of it :-)
-Dan
> Forum: Cfengine Help
> Subject: Looking for suggestions on ordering
> Author: Authority
> Link to topic: https://cfengine.com/forum/read.php?3,17628,1
The error says: "Redefinition of body start for location". So clearly you must
have a "body location start {..." somewhere else in your promises! You can
only have ONE body of a type with a certain name, and cfengine_stdlib.cf is
defining the second one in your collection of promises...
-Dan
No, this is not a Nova-ony feature. The policy_server.dat file is part of
the Nova configs, but you can build the functionality yourself in Community.
-Dan
> Ummpossibly. :) I've never used Community, so I guess I don't know exactl
> y where the differences are. However, promises.cf reads th
With Mark's permission, I am wearing my USENIX hat for this email and
promoting a couple of upcoming events that will prominently feature
Cfengine...
First, I'd like to make sure you are all aware of the Configuration
Management Summit in Boston on June 24 (details are at
http://www.usenix.org
I hear that you would prefer to write it one way, but sorry, you can't. Well,
you sort of can, you need to look at "policy => free" or the synonym for this
which is "policy => overridable", but we think you'd be wrong.
Let me give you a counterexample. I program in C, and there really have been
Take a look at methods promises, and specifically the value given to the
usebundle attribute.
http://www.cfengine.org/manuals/cf3-reference.html#methods-in-agent-promises
-Dan
> Hello. I need some feature like dependecies in bundles.
>
> For example: i have 5 agent bundles that can do specific
Is one of the commands perhaps to create the directory you want? Why not use
the "create => true" attribute and just let the promises converge to the
desired state?
-Dan
> --===1770191284==
> Content-Type: multipart/alternative; boundary=0016e6470cf819ef39048813b6ae
>
> --0016e6470c
Justin-
Do you really have three separate licenses, or do you just want to distribute
policy from three servers? If the latter, the host_licenses_paid is the
number of licenses, not the number of suppoorted clients (and so you don't
need to have a conditional control parameter :-) And if the for
anged mine to an agent bundle and it worked fine, but I'd like to do
> it in a common bundle for global access. Is this a technical limitation
> with how common bundles work?
>
> Justin
>
> -Original Message-
> From: Daniel V Klein [mailto:d...@lonewolf.com]
&
1 - 100 of 114 matches
Mail list logo
