!suse will match a redhat machine. So that over rides the previous redhat
declaration. I don't know what your goal of the ! assertion is but you
might try !suse.!redhat.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the
Will existing DBs be converted as Cf moves to a new DB format?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the in
What will happend to existing data once the backend DB changes?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the i
Did you try escaping $ instead? Give us a selfcontained bundle to compare
with your results.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential inf
In trying to make my policy more modular and easier to make changes I've
developed what you see below (might be helpful to others). This nice
thing about this is that adding or changing files can all be done in vars
without having to touch the actual promise.The one problem is handling
the
Consider this example.
body common control {
bundlesequence => { "test" };
}
bundle common g {
classes:
"all" expression => "any";
}
bundle agent test {
vars:
sunos_5_10::
"ltwo" slist => { "three", "four", "five" };
redhat::
"lone"
Mark Burgess wrote on 2009-12-01 15:46:59:
>
> If the symbol is not defined, it cannot be expanded
Is there a way to append an existing list?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee
Mark Burgess wrote on 2009-12-02 00:21:41:
>
> Exactly as you did. It works for me.
What I did I would consider concatenating lists. I was hoping to append
as in:
vars:
any::
"x" slist => { "one", "two", "three" };
redhat::
"x" slist => { "@{x}",
I do not believe cf-monitor is required in order for cf-agent and
cf-serverd to work properly.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential in
Greetings,
I need some help with a editing promise that uses list iteration.
body common control {
bundlesequence => { "test" };
}
bundle common g {
classes:
"all" expression => "any";
}
bundle agent test {
vars:
any::
"x" slist => { "one", "two", "thr
Has anyone done any work gathering the status of services via the Solaris
SMF tool kit? There are some interesting caveats.
1. svcs returns zero whether the service is running or not.
2. Service names have unusual characters (:/) that can update Cf if
trying to make them into classes.
Sincere
Thanks for the tip Mark. Using this method again for something different
get's me some unexpected results:
body common control {
bundlesequence => { "foo" };
}
bundle common g {
vars:
sunos_5_10::
"svc_cmd" string => "/usr/sbin/svcadm",
help-cfengine-boun...@cfengine.org wrote on 2009-12-02 16:13:56:
> Hi,
>
> Why don't you use only ${d} in methods?
>
> methods:
> "any" usebundle => subfoo("${d}");
>
> I think you would get expected results.
It does but as Mark pointed out it is a little more efficient to pass the
list one
Anyone?
Sincerely,
--
Neil Watson
416-673-3465
help-cfengine-boun...@cfengine.org wrote on 2009-12-01 21:07:40:
> I've put together a test bundle for this. Nova gives out these errors:
> cf-promises -f ./foo.cf
> Invalid mode string (${mod})
> Problem validating a mode string
> Promise (version
Could some folks kindly take a moment to attempt to duplicate my results
using either Nova or community?
Sincerely,
--
Neil Watson
416-673-3465
Mark Burgess wrote on 2009-12-03 00:38:26:
>
> Variables are not getting expanded / not defined. Hard to say more.
>
> nwat...@symcor.com wrote:
> >
It is attached to the thread parent.
https://cfengine.org/pipermail/help-cfengine/2009-December/006732.html
Sincerely,
--
Neil Watson
416-673-3465
Mark Burgess
2009-12-03 09:09
To
nwat...@symcor.com
cc
help-cfengine@cfengine.org
Subject
Re: invalid mode string error.
Then you need to
Does anyone have a working method for building cf3 on aix that they can
share? For me configure seems for work but make step returns an error:
/bin/sh ../ylwrap cf3parse.y y.tab.c cf3parse.c y.tab.h
cf3parse.h y.output cf3parse.output -- bison -y -d
got /home/nwatson/cfengine-3\.0\.2
Nicolas Charles wrote on 2009-12-03 09:20:07:
> With the latest svn build, on Ubuntu, seems to work find.
> Here is the result :
> # cf-agent -Kf ./foo.cf
> R: ALARM: /etc/vim/vimrc wrong contents or perms 0644, root, root
> R: /etc/vim/vimrc correct contents and perms 0644, root, root
>
I get
Jean-Noel,
I'm not certain what the bug in question is. It would be helpful if you
posted, along with your test case, your expected and actual results.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
a
So Nova reports errors and community does not. Mark, does Nova perform
some stricter parsing that would account for this?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain propri
I have a promise that used tidy but in a warn only manner. In spite of
this files are still being deleted.
##
body common control {
bundlesequence => { "test" };
}
bundle common g{
vars:
"masterfiles" string => "/var/cf-masterfiles";
"inputs" string
I: Comment:
Test case:
##
body common control {
bundlesequence => { "test" };
}
bundle common g {
classes:
"all" expression => "any";
}
bundle agent test {
files:
any::
"/tmp/foo"
You might need the openssl-devel package to be installed.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the intende
If a promise is linking warn actions are also ignored and links are
created.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
File /tmp/foo is owned by [root], group [bin]
I: Made in version 'not specified' of './bar.cf' near line 18
!! File /tmp/foo is owned by [root], group [bin]
I: Made in version 'not specified' of './bar.cf' near line 18
group is wrong:
[r...@unxxhd01 inputs]#
Try escaping \...@.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the intended recipient, any use, copying, disclo
Eystein Måløy Stenberg wrote on 2009-12-09
03:15:13:
> Hi Neil,
>
> I think the reason for this is that the correct mode-string is always
> a scalar, a short string, while the correct owners/groups can
> potentially be long lists.
> Printing valid owners/groups could thus lead to messy output.
I've made good progress is producing reusable policies with CF3. Consider
a security policy that checks the permissions of a motley group of files.
At first one might consider setting a a list or two and going from there.
vars:
"644-mode" string => "0644";
"644-usr" string
Thanks for the feedback Ed. It is possible to try to further hide some
repeated information such as owners and modes. However this value still
has to be set somewhere. If it is not in the current location as I have
shown you'll have to look in multiple places to find all the information.
Let
Mark Burgess wrote on 2009-12-09 15:25:34:
>
> For me the main thing has been to provide mechanisms - take them or
> leave them - for
> expressing complex patterns in different ways. You pick the approach
> that floats your
> particular boat.
tmtowtdi (
http://en.wikipedia.org/wiki/There%27s_mo
It's too early for me to tell what the best methods of code style will be
for CF. Experience tells me that as policies grow, are deployed to more
hosts and are managed by multiple people such separation of data and
function are very helpful. If others have different styles that they
think are
help-cfengine-boun...@cfengine.org wrote on 2009-12-10 12:09:57:
> Christopher Browne wrote:
> > - In reflecting on it, I think I'd like for some of the derived
classes
> > to be derived out of data residing in external sources like
LDAP/DBMS.
>
> Yes, a native ability to import, AND export,
Using community I do not get these warnings. Have you tried my example in
community or Nova?
Sincerely,
--
Neil Watson
416-673-3465
help-cfengine-boun...@cfengine.org wrote on 2009-12-13 17:13:42:
> I do this all the time and never get any warnings.
>
> On Fri, Nov 27, 2009 at 8:47 PM, Neil W
Brendan Strejcek wrote on 2009-12-13 16:14:34:
> Thinking about this slightly on a slightly higher-level:
>
> If you are implementing organizational processes similar to the
> Visible Ops recommendations, you might want to think setting up a
> modeful policy; that is, support Cfengine running in
I built cf2 with esx 2.x some years ago. At that time esx was simply
repackaged Red Hat.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential informa
I found the cf3 learning curve to be steep. Indeed the more I learn the
less I seem to know. The tough syntax is why I wrote my own tutorial.
http://watson-wilson.ca/blog/cf3-tutorial.html
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attach
Do you have another bundle or body part elsewhere named 'start'? If not
try to make an entirely self contained example and see if it exhibits the
same behaviour.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive
Mark Burgess wrote on 2009-12-15 08:59:37:
> The syntax is not tough. It is incredibly simple. One pattern to
> follow for everything.
> You mean that the idea is unfamiliar and the scope is huge. That's
different.
Perhaps. Originally I found the examples were not practical enough for me
to
Did the configure come out alright? Did if find openssl, pcre and
Berkeley DB?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If y
Do you need both the i386 and x84_64 libs? Perhaps remove one.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the i
Can you tell us why you need to know the IP address? There might be
another way to meet that end result.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or con
Verbose output might be better than debug output for discovering faults.
Also it might be more efficient to have CF copy the files recursively than
copying a tar ball and then extracting it. You'll have greater
reliability too as CF will log any problems during the copy.
Sincerely,
--
Neil Wat
ion => warn_now,
perms => system(
"0600",
"nwatson",
"bin"
),
classes => cdefine(
"/tmp/foo_kept",
"/tmp/foo_repaired",
We've all been or are still there. CF's promise theory offers a different
way of thinking. It's takes some time to adjust to it. Think of it like
calculus. For a while your brain just explodes but, eventually it starts
to make sense.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY
Brendan Strejcek wrote on 2009-12-17 10:44:35:
> This is reliable, but not very efficient. It requires at least 2N file
> checksum calculations (where N is the number of files in the tree
> being copied or verified) on every pass. You are also using 2N space,
> since you are storing a copy in ${s
You can pass the entire list for greater efficiency.
...
files:
redhat|suse|sunos_5_10::
"${crontabs}/root"
comment => "Added cron jobs to table.",
edit_line => AppendIfNoLine("@{cron.root_cron_lines}");
}
bundle edit_line AppendIfNoLine(l
See this earlier thread.
https://cfengine.org/pipermail/help-cfengine/2009-December/006742.html
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential in
If I understand you correctly you have a problem with two masters. CF is
updating binaries but so is the Solaris package manager. Only one service
should be in charge of keeping CF binaries up to date. Either is fine but
be consistent.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALIT
http://www.cfengine.org/manuals/cf3-reference.html#copy_005ffrom-in-files
'preserve'
You are correct however that trusting the perms on the source file is not
ideal.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclus
The server probably broadcasts using its Ethernet address not localhost.
So the agent will look for the server's public key to match. You can also
run the server non forked in verbose to see interesting output.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication
6: warning: passing arg 4 of pointer to function makes
pointer from integer without a cast
instrumentation.c:426: error: too few arguments to function
gmake[2]: *** [libpromises_la-instrumentation.lo] Error 1
gmake[2]: Leaving directory `/home/nwatson/src/cfengine-3.0.2/src'
gmake[1]: *** [all]
Odd message from cf-promises. I'm using community 3.0.2. 3.0.3 is later
in my agenda.
nwat...@tofrcpl58903 ~/cfengine/config/branches/sandbox-nwatson/inputs $
cf-promises.exe -f ./promises.cf
cf3:./cfengine_stdlib.cf:722,53: Constraint lvalue package_multiline_start
is not allow
I built and tested with 3.0.3 and the 'package_multiline_start' error
disappeared. I assume that this is a new feature. Thus I suggest that
cfengine_stdlib.cf be versioned so that we can be assured compatability.
I did still get this error even with 3.0.3:
Validation: Error reading assumed int
Your example does not make sense. The agent calls a bundle
'replace_lines'. The bundle in your example is called 'insert_lines'.
Please try to provide a self contained example along with expected and
actual results.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This commun
The reference states that "You should never put cf-agentor cf-agent
directly inside cron without the cf-execd wrapper."[sic] I'm not certain
that I see the logic in this. In my policies I have a processes promise
that ensures that cf-execd is running. As a backup also like to have a
crontab
I'm testing a solution that runs cf-execd -F through cron, as in the
reference. My processes promise for cf-execd takes advantage of PCRE's
negative lookahead and does "cf-execd(?! -F)" thus is looks for any
cf-execd process that is not followed by ' -F'. So far this is working
well.
Sincere
A 'push' is done the same way in both versions of Cfengine. In cf2 one
uses cf-run. In cf3 one uses cf-runagent.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary an
I've never seen this before. What does it mean?
/var/cfengine/cfengine_lock_db: page 15: reference count overflow
PANIC: Invalid argument
PANIC: fatal region error detected; run recovery
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachmen
Happy New Year everyone,
While working with the new standard library I notice two things. First,
there lacks a body to change just ownership. There is one for mode, mode
and user, mode user and group, and finally just user. So adding this
would be helpful:
body perms og(u,g) {
owners =>
Yes. You can do this. Here is a recipe with CF version 2. You'll have
to figure out the Cf 3 way but the principle is the same.
http://watson-wilson.ca/blog/cfcookbook.html#SECTION00020
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, includ
Using cf3 you can keep the list of hosts in a separate file and read them
into a list(
http://www.cfengine.org/manuals/cf3-reference.html#Function-readstringlist
). This could make thing a little more manageable. Still, I agree that
if you are trying do to this for thousands of hosts a more ded
Anyone?
Sincerely,
--
Neil Watson
416-673-3465
help-cfengine-boun...@cfengine.org wrote on 2009-12-31 14:36:23:
> Greetings,
>
> In discussing file promises it is suggested in the reference, and on
> this list, that sometimes it is desirable to have separate promises to
> mind the contents and
I'm running Community 3.0.3.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the intended recipient, any use, copying
Do you have inform set somewhere?
http://www.cfengine.org/manuals/cf3-reference.html#inform-in-agent
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidenti
I think that purge only works with recursive copies.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the intended rec
To me the promise says make sure this file is present on the client. If
the source file goes missing I'm not sure it is wise to have it
automatically removed from the client.
Sincerely,
--
Neil Watson
416-673-3465
Tom Van de Velde
2010-01-08 08:52
To
nwat...@symcor.com
cc
help-cfengine@cf
I believe that you can do this now. It works very similar to the expand
variables method used in cf2.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confide
My point is that the promise is still changing permissions even though no
permissions are specifically promised. It seems that the agent defaults
to setting the target permissions to match the source permissions.
Sincerely,
--
Neil Watson
416-673-3465
help-cfengine-boun...@cfengine.org wrote o
The 'servers' option in the copy_from body takes an array of servers.
http://www.cfengine.org/manuals/cf3-reference.html#copy_005ffrom-in-files
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee a
I was teaching him to fish :) Seriously if anyone uses multiple servers
in production and has tested failover I'd love to hear about your
experience.
Sincerely,
--
Neil Watson
416-673-3465
Mark Burgess wrote on 2010-01-13 14:15:07:
>
> To follow up from Neil, try setting localhost as your f
'define' is the name of your classes body compound (which you probably
don't have or define is a reserved word). It is disuccsed in the
reference[1]. There are probably some predefined classes bodies in the
standard library[2]
1.
http://www.cfengine.org/manuals/cf3-reference.html#classes-in-
help-cfengine-boun...@cfengine.org wrote on 2010-01-14 14:38:18:
>
> commands:
> sendmailmake::
> "cd /etc/mail";
> "/usr/bin/make";
> "/sbin/service sendmail restart";
I'm wary that these commands might not always go in order.
Sincerely,
--
Neil Watson
416-
Mark briefly talks about ordering at the end of this document:
http://www.cfengine.org/pages/compare
If you register at cfengine.com and login there is an command validator
under tools that might help with using ';'.
The 'contain' body part will allow you to 'chdir'. That will save you the
cd c
If you run the agent -v and pipe to less you'll see the verbose output.
Look at where the file promise is processed. What is happening?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may
Can we see the code? Perferably in a selfcontained example.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the inte
/sbin/rc3.d/S998cf-serverd is a strange path to me. What does ls -l
/sbin/rc3.d/S998cf-serverd return?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confid
Anyone?
Sincerely,
--
Neil Watson
416-673-3465
help-cfengine-boun...@cfengine.org wrote on 2010-01-15 19:59:44:
> Greetings,
>
> I have a question about setting ifelapsed to 0 for certain promises.
> Consider this code. The goal is to repair a service config file. If
> this happens stop the s
I've compiled cfengine on solaris with some libs in a custom location.
When I run cf it fails unless I set LD_LIBRARY_PATH to point to the custom
libs. Is there a way to build the cf binaries with a built in default
library search path?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY
What about setting links=follow ?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the intended recipient, any use, co
You can do this. I've done it myself. However, you can overload yourself
with useless alerts. e.g. ALERT: Cfengine doing it's job! Tongue in
cheek to be sure but you get the point. It is however not a bad idea to
log the repair, quietly.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENT
Something like this:
files:
"${trg}"
perms => m( "${mod}" ),
action => warn_now, # Passive only at this point.
classes => cdefine(
"${trg}_mode_kept",
"${trg}_mode_repaired",
"${trg}_mode_failed"
If you are new to cfengine and not inheriting an existing install then you
should be studying version 3. It is the future.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain propr
Please consider this test.
body common control {
bundlesequence => { "test" };
}
bundle common g{
classes:
"all" expression => "any";
}
bundle agent test {
commands:
any::
"/bin/echo \'hello world\'",
action => immediate;
}
body action imm
Has the agent been told to import the file 'cf.syslog-ng'?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the intend
What is the expected behaviour when performing a recursive copy where the
source contains symlinks?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confident
My promise is to copy a lib directory. Inside are links pointing to files
in the same directory. Should the copy preserve these links? Initial
testing indicates that the links are ignored. No options for copy_from
seem related to me.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY
Currently I have removed just a link and it is not be recreated by the
copy. Is it possible that links are being followed instead?
Sincerely,
--
Neil Watson
416-673-3465
Mark Burgess wrote on 2010-01-21 11:09:30:
>
> I would expect these to be preserved, but the outcome might be order
> dep
Sorry Mark, my mistake. My recurse depth was not deep enough to catch the
missing link.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential informat
Consider this:
bash-3.00# ldd /var/cfengine/bin/cf-key
libdb-4.8.so => /var/cfengine/usr/BerkeleyDB.4.8/lib/libdb-4.8.so
libcrypto.so.0.9.7 =>/usr/sfw/lib/libcrypto.so.0.9.7
libpthread.so.1 => /lib/libpthread.so.1
libpq.so.4 =>/usr/lib/libpq.so.4
I am building community 3.0.2. I don't need it yet the build is linking
to it. This is hampering my attempts at a more portable build. How can
disable this at configure time? I tried --disable-mysqlclient and a few
other combinations but configure does not like the options.
Sincerely,
--
Ne
Using --without-sql seems to work. Could this lead to any other problems?
Sincerely,
--
Neil Watson
416-673-3465
Mark Burgess
2010-01-21 15:03
To
nwat...@symcor.com
cc
Nakarin Phooripoom , help-cfengine@cfengine.org
Subject
Re: odd lib dependency
You could compile on a machine that d
Greetings,
Here are some things I'd like to see in future versions of Cf3.
1. A tool or command option to confirm an authentication handshake from
the client side. Currently one is forced to run the agent in verbose mode
and search though the output for the H.A.I.L sections. It need not be
Can you get allclasses.txt without needing some form of inputs? When I'm
building or bootstrapping new architecture clients I often don't have
fully prepared inputs ready. This is where handshake tests and hard class
outputs are very valuable.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDEN
I'm running 3.0.2 community. I have a bundlesequence list. A comma at
the end causes cf-promises to report an error.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietar
I can confirm that a comma on the last item of a list is no longer a
syntax error. In regards to using cf-runagent for testing handshakes I
can only see that possible if the run agent now be targeted to a single
host. My testing Community 3.0.2 indicated that cf-runagent always went
through a
I favour the idea of keeping apt-get update and apt-get upgrade separate.
In this manner I can schedule an update and know that subsequent upgrades
for all hosts will match the update schedule. This keeps your upgrades
all on the same time line.
Sincerely,
--
Neil Watson
416-673-3465
CONFIDE
I hardly use disk quotas myself but, if they were managed by Cfengine I
think I'd be more inclined. Thoughts?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or
help-cfengine-boun...@cfengine.org wrote on 2010-01-26 18:14:56:
>
> # "Live" .bashrc
> { home/.bashrc
> AppendIfNoSuchLine ". /etc/foo"
> }
What does verbose output say? I'm guessing that this promise will be
ignored since, being a relative path, Cf cannot determine where t
Running Community 3.0.2 on Redhat 5.3, 32 bit.
cf-promises -x in gdb
Start was not at 31 -> 47
-> BlockTextMatch - ok
5. Testing promise attribute completeness
[New Thread 0xb7f3c6d0 (LWP 27088)]
Program received signal SIGSEGV, Segmentation fault.
0x08086dcb in cfPS (level=cf_error, status=119
root squash issue?
Sincerely,
--
Neil Watson
416-673-3465
CONFIDENTIALITY WARNING
This communication, including any attachments, is for the exclusive use of
addressee and may contain proprietary and/or confidential information. If you
are not the intended recipient, any use, copying, disclosu
1 - 100 of 238 matches
Mail list logo
