Hello,
Iterations do not behave as expected for vars sections, in my opinion
at least, because vars section seems to be evaluated only once.
- The manual
cf3 Reference Manual explains (2. A simple crash course in concepts /
2.8 Normal ordering) :
"Within a bundle, the promise types are executed
The quick answer is "you don't".
The longer answer is: an slist (or ilist or rlist) can only be iterated over
in it's entirety by referencing the list as a scalar (and then cfengine
iterates over the whole list automagically). You can index an array (which
is what 'foo' is in your example), but a
On 5 Mar 2010, at 7:47 am, Bas van der Vlies wrote:
>
> On Mar 5, 2010, at 12:52 AM, Jessica Greer wrote:
>
>> Thanks to everyone for the responses - that's helpful. Tim, I
>> especially like your idea of mailing out diffs, so everyone stays in
>> the loop, and teammates don't have to aler
Because it is a declarative promise language - which is somewhat different
than the usual imperative programming language model. A fine point, but
different nontheless. Declarative languages are hard to get your head around
if you think in purely imperative terms. Once you "let go" and allow
you
Hello,
Jonathan Clarke and I will attend and make a presentation about cfe (at
least, we hope it'll be accepted).
By the way, is anybody familiar with what they expect ? I guess it's a
rather technical presentation that is expected, but I'd like to be sure
about it...
Regards,
Nicolas
Mark Bu
Great news !
The intended audience is : sysadmins.
And quoting one of the organisators :
"it's not like fosdem where we want the core developers of a project to
present ... it's more about what/ how/why and using it ... :)"
Cheers,
f.
On 5/03/2010 12:09, Nicolas Charles wrote:
> Hello,
>
> Jo
On 05-03-10 10:30, Tim Cutts wrote:
>
> On 5 Mar 2010, at 7:47 am, Bas van der Vlies wrote:
>
>>
>> On Mar 5, 2010, at 12:52 AM, Jessica Greer wrote:
>>
>>> Thanks to everyone for the responses - that's helpful. Tim, I
>>> especially like your idea of mailing out diffs, so everyone stays in
>>> th
I had the same issue with 3.0.2. I had to move my class declaration in a
common bundle in order to make it work.
Marco
-Original Message-
From: help-cfengine-boun...@cfengine.org
[mailto:help-cfengine-boun...@cfengine.org] On Behalf Of lefollfr
Sent: Friday, March 05, 2010 3:34 AM
To:
Hello Marco,
I tested with 3.0.3 and SVN revision 870.
With classes defined somewhere earlier in the bundlesequence, i.e.
already defined during first and single evaluation of a vars section,
everything works fine indeed. But I think it breaks the "declarative"
approach of promises, and requires a
Frederic,
All I was saying is that I did come across the same issue and I came up with a
work around. That being said I am of your opinion that if a local class gets
'activated' at any point in time then the vars that are declare base on the
existence of that class should become visible at tha
daniel.kl...@cfengine.com wrote:
> Because it is a declarative promise language - which is somewhat different
> than the usual imperative programming language model. A fine point, but
> different nontheless. Declarative languages are hard to get your head around
> if you think in purely imperativ
CF does what it does and does it well whether it is programming language
or not. It is definitely not like a scripting language and because of
this I often find I have to alter my approach to promises. My mind tries
to rely on its experience with shell and perl scripts but those methods
don't
I'd like to throw out my own question about version control. I've
finally finished my initial policy files for our pilot deploy (for the
most part; need to update comments). There are four files that we
currently maintain/edit: promises.cf, dg.config.cf, dg.library.cf, and
dg.bundles.cf. The rest o
Thanks, Nicolas. I understand Cfengine 3 CAN read Cfengine 2 policies
- the feature is "Auto-encapsulation of Cfengine 2 for incremental
upgrade", see at the bottom of http://cfengine.com/pages/cfengine3
I just want to find out HOW to do it. :) I am ready for an
incremental upgrade of a product
I prefer to have the whole lot in version control. This makes it your go
to place for disaster recovery or migration. Sometime you may want to use
multiple repositories.
/var/cfmasterfiles/config (general cfengine files)
/var/cfmasterfiles/dns (hostmaster Bind files)
/var/cfmasterfiles/app (a
Aleksey, Cfengine 3 provides the possibility for cfengine 3 and cfengine 2 to
coexist
during an upgrade phase. It cannot *read* Cfengine 2 policies. Cfengine (the
company) can
help transform a configuration.
Aleksey Tsalolikhin wrote:
> Thanks, Nicolas. I understand Cfengine 3 CAN read Cfeng
Greetings,
I see that cf-3.0.4 is out there, but access if forbidden to download it. I may
have jumped the gun, but I am very excited.
I haven't been to the cfengine web page lately, but, Wow! Lots of work and new
stuff. Very impressive. Did ya happen to see Jarle Bjorgeengen's cfengine vs
pup
Thanks Matt, sorry for the lapse.
Matt Richards wrote:
> Greetings,
>
> I see that cf-3.0.4 is out there, but access if forbidden to download it. I
> may have jumped the gun, but I am very excited.
>
> I haven't been to the cfengine web page lately, but, Wow! Lots of work and
> new stuff. Ver
I just wanted to throw out a couple of thoughts about copy_from that
have come up in my mind during my policy development:
* The reference manual doesn't specify the default value for copy_from's
copy_backup attribute. (That goes for a number of things in the
documentation, actually.)
* cfengine_
On Fri, Mar 05, 2010 at 02:11:42PM -0500, Justin Lloyd wrote:
>I just wanted to throw out a couple of thoughts about copy_from that
>have come up in my mind during my policy development:
Agreed on all three points.
I'd like to add one more suggestion as well:
* Include the ability to call expand
help-cfengine-boun...@cfengine.org wrote on 2010-03-05 14:37:57:
> I'd like to add one more suggestion as well:
>
> * Include the ability to call expand_scalars in the same promise as
> copying a file from a remote host. Currently, this has to be done
via
> two separate promises: one to
On Fri, Mar 05, 2010 at 02:45:19PM -0500, nwat...@symcor.com wrote:
>help-cfengine-boun...@cfengine.org wrote on 2010-03-05 14:37:57:
>> I'd like to add one more suggestion as well:
>>
>> * Include the ability to call expand_scalars in the same promise as
>> copying a file from a remote host.
I think I understand Jesse's rationale but I agree with Neil.
Analogously to the Unix command philosophy, IMHO, bodies (and often
bundles) should do one thing and do it well. Use multiple promises in
conjunction like you would use command-line pipes.
Jesse, have you maybe tried something with a me
I can compile fine with no options but as soon as I get clever I break
something:
./configure --prefix=/home/nwatson/cfengine
--with-pcre=/home/nwatson/cfengine
--with-berkeleydb=/home/nwatson/cfengine --without-sql
--sbindir=/home/nwatson/cfengine/bin
Configure looked good:
checking for Berk
On Fri, Mar 05, 2010 at 03:07:36PM -0500, Justin Lloyd wrote:
>I think I understand Jesse's rationale but I agree with Neil.
>Analogously to the Unix command philosophy, IMHO, bodies (and often
>bundles) should do one thing and do it well. Use multiple promises in
>conjunction like you would use co
Jesse,
Perhaps you should take a fresh approach. Restate your original goal that
started all your template efforts. Perhaps someone else has another way
to do it.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclus
Hi,
I'm using community version 3.0.3 and I'm getting the following error:
cf3:./test.cf:34,96: Unknown built-in function regextract(), near token ')'
Is there another way of extracting values into strings with 3.0.3? I'm hoping I
don't have to upgrade all the servers to 3.0.4 yet.
This is wh
According to the documentation, the following is not convergent:
bundle agent ensure_pam_passwords_clear {
files:
"/etc/ldap.conf"
edit_line => pam_password("clear");
}
bundle edit_line pam_password(method) {
replace_patterns:
# haven't tested whether PCRE zero
Never mind, I solved my own problem. After I wrote it up and sent it, I
started an email to my team about this issue, and it was then the
solution dawned on me. To avoid the convergence issue, I just needed a
class to restrict when the non-convergent replacement could happen by
first reading the fi
It seems like the configure script could not find db header files. A good one
should look like;
checking for BerkeleyDB location in /var/cfengine.community...
/var/cfengine.community
checking Berkeley DB API... 4.8.24 OK
checking for db_create in -ldb... yes
You might need to do ldconfig or pos
Ok, take 3. It turns out, due to how Cfengine does changes in memory and
compares that to the on-disk file, I don't even need to do the class
check. Just comment all of the pam_password lines out and then do the
uncomment_first_occurrence. The two replace_patterns promises together
ARE convergent.
Thanks, Mark.
What is meant by "auto-encapsulation", please?
Is it that I can I run cf-execd and cfexecd out of crontab and they'll
happily share the /var/cfengine work directory?
Thanks!
Aleksey
On Fri, Mar 5, 2010 at 10:34 AM, Mark Burgess wrote:
>
> Aleksey, Cfengine 3 provides the possibi
Forum: Cfeninge Help (Mailing list)
Subject: Re: Iterations for convergence : vars do not iterate
Author: mark
Link to topic: https://cfengine.com/forum/read.php?3,7774,7788#msg-7788
I suggest you report it as a bug on the bug tracker with a reproducible example
of what you mean
33 matches
Mail list logo
