Xymon 4.3.22-beta Release Download 🔗 link

Hello all!

Xymon-4.3.22-beta has been posted to Sourceforge, containing bug fixes and
a smattering of new features and improvements since 4.3.21. It's reachable
at
http://sourceforge.net/projects/xymon/files/Xymon/4.3.22/xymon-4.3.22-beta.tar.gz/download

Some notable changes are at
https://sourceforge.net/projects/xymon/files/Xymon/4.3.22/, while the full
Changelog is at
https://sourceforge.net/p/xymon/code/7707/tree/branches/4.x-master/Changes


I know there are a few pending bug fixes already that I'd like to get in
before the next release (multigraphs, if I can track down the source),
however the number of features warrants some testing. Please let me know
if you notice any issues (especially on non-Linux systems, as I don't have
as much direct access to those).

In particular, the changes in how xymonnet interprets 301/401/403 HTTP
status codes could cause some confusion. Those who've customized their
NTPDATE or TRACEROUTE command lines should check out the new Options
variables. And finally, it would be helpful to have high volume
environments (>60000 hosts) testing the "noflap" option for any notable
performance impact. (It should be very slight.)


If all goes okay, I'd like to release the final 4.3.22 some time next week.


Regards,

-jc

J.C. Cleaver wrote:

I never got any feedback regarding the chpasswd patch, it adds further 
(unpublished) functionality similar to useradm.cgi, we have need of this 
  and would appreciate it if could be added to the main distribution.

Thanks
-- 
Andy

J.C. Cleaver wrote:

Apart from user-ce4d79d99bab@xymon.invalid who could not repeat my problem with 
integer arithmetic in combostatus, I never saw any feedback about the 
attached patch which fixes our problem on Linux builds.

As a reminder, this rule in combo.cfg :-

angelica.shadymint.com.combo=(angelica.shadymint.com.http + 
angelica.shadymint.com.bbd + angelica.shadymint.com.procs)>=3

is calculated incorrectly on Linux, but strangely is OK on Solaris.

The error message we get on Linux is :-

2015-10-13 00:39:07.263726 Item '3' has no testname part
2015-10-13 00:39:07.263762 Item '3' has no testname part
2015-10-13 00:39:07.263769 Invalid data for symbol calculation - missing 
host/testname: 3
6736 2015-10-13 00:39:07.263777 Symbolic 
'(angelica.shadymint.com.http+angelica.shadymint.com.bbd+angelica.shadymint.com.procs)>=3' 
converted to '(1+1+1)>=0'

You will not know from the web page that this has failed because 
'(1+1+1)>=0' is still true, but its not the answer we need to find.

Thanks
-- 
Andy

Hi J.C.,

On Thu, Oct 29, 2015 at 06:44:07PM -0700, J.C. Cleaver wrote:

Builds fine on Debian Unstable. And I could drop tons of patches from
the package, yay! :-) Thanks!

		Kind regards, Axel Beckert
-- 
Axel Beckert <user-96d9963fe797@xymon.invalid>       support: +41 44 633 26 68
IT Services Group, HPT H 6                  voice: +41 44 633 41 89
Departement of Physics, ETH Zurich
CH-8093 Zurich, Switzerland		   http://nic.phys.ethz.ch/

On 10/30/2015 3:36 AM, Andy Smith wrote:

Sorry for this miss; I intended to get this one in. It'll be in the full 4.3.22 release.

-jc

On 10/29/2015 5:44 PM, J.C. Cleaver wrote:

These look like important corrections and additions. I particularly will like the ability to disable "flap detection". I have been out-of-office and have not yet had time to test this release. I hope to make time today or tomorrow to do so.

I am concerned that I see no mention in the changelog of any modification of combo test behavior. The behavior of combo tests (as defined in combo.cfg) is badly broken under Solaris. Failure, test conditions, and a stack excerpt are mentioned in my email to the list:
http://lists.xymon.com/archive/2015-October/042382.html

It works correctly on my CentOS installation, but not Solaris. I was able to move my combo test to the CentOS server and forward the result to my Solaris server. This gets the job done for me, but should I be hit by a truck it could be a little difficult for the next person to unwind the convoluted path.

-- 
    Do things because you should, not just because you can.

John Thurston    XXX-XXX-XXXX
user-ce4d79d99bab@xymon.invalid
Enterprise Technology Services
Department of Administration
State of Alaska

On Mon, November 2, 2015 10:21 am, John Thurston wrote:

The general issue with parentheses should be fixed as of
https://sourceforge.net/p/xymon/code/7708/, but more testing in that
regard would always help.

I think the Solaris problem here is probably our good friend
NULL-into-printf again, though I'd need to see the page output from the
CentOS side in comparison to really say for sure.

I hadn't even announced the RC2 tarball from this morning yet, but if you
get a chance if you could try the attached patch here with it on your
Solaris machine, I believe it will solve the core issue.

combonull.patch
http://sourceforge.net/projects/xymon/files/Xymon/4.3.22/xymon-4.3.22-rc2.tar.gz/download


Regards,

-jc

John Thurston wrote:

Just built 4.3.22 on my Solaris 10 with gcc version 3.4.6

bash-3.00$ uname -a
SunOS oregano 5.10 Generic_127111-06 sun4u sparc SUNW,Ultra-60

Repeating your 4 test cases, these all work for me :-

oregano.shadymint.com.combo= oregano.shadymint.com.http && oregano.shadymint.com.bbd

bash-3.00$ combostatus --no-update --debug
22025 2015-11-02 20:57:25.333972 Opening file /apps/xymon/server/etc/combo.cfg
22025 2015-11-02 20:57:25.339010 Symbolic 'oregano.shadymint.com.http&&oregano.shadymint.com.bbd' converted to '1&&1'
22025 2015-11-02 20:57:25.339322 Adding to combo msg: status oregano,shadymint,com.combo green Mon Nov  2 20:57:25 2015


oregano.shadymint.com.combo= oregano.shadymint.com.http && (oregano.shadymint.com.bbd==0)

bash-3.00$ combostatus --no-update --debug
22102 2015-11-02 20:59:13.388510 Opening file /apps/xymon/server/etc/combo.cfg
22102 2015-11-02 20:59:13.393672 Symbolic 'oregano.shadymint.com.http&&(oregano.shadymint.com.bbd==0)' converted to '1&&(1==0)'
22102 2015-11-02 20:59:13.393944 Adding to combo msg: status oregano,shadymint,com.combo red Mon Nov  2 20:59:13 2015


oregano.shadymint.com.combo=  (oregano.shadymint.com.bbd==0) && oregano.shadymint.com.http

bash-3.00$ combostatus --no-update --debug
22506 2015-11-02 21:00:35.628811 Opening file /apps/xymon/server/etc/combo.cfg
22506 2015-11-02 21:00:35.633988 Symbolic '(oregano.shadymint.com.bbd==0)&&oregano.shadymint.com.http' converted to '(1==0)&&1'
22506 2015-11-02 21:00:35.634350 Adding to combo msg: status oregano,shadymint,com.combo red Mon Nov  2 21:00:35 2015


oregano.shadymint.com.combo= (oregano.shadymint.com.http && (oregano.shadymint. com.bbd==0))

bash-3.00$ combostatus --no-update --debug
22708 2015-11-02 21:01:56.237602 Opening file /apps/xymon/server/etc/combo.cfg
22708 2015-11-02 21:01:56.245084 Symbolic '(oregano.shadymint.com.http&&(oregano.shadymint.com.bbd==0))' converted to '(1&&(1==0))'
22708 2015-11-02 21:01:56.245326 Adding to combo msg: status oregano,shadymint,com.combo red Mon Nov  2 21:01:56 2015


oregano.shadymint.com.combo=  ((oregano.shadymint.com.bbd==0))

bash-3.00$ combostatus --no-update --debug
22752 2015-11-02 21:03:11.950561 Opening file /apps/xymon/server/etc/combo.cfg
22752 2015-11-02 21:03:11.958343 Symbolic '((oregano.shadymint.com.bbd==0))' converted to '((1==0))'
22752 2015-11-02 21:03:11.958782 Adding to combo msg: status oregano,shadymint,com.combo red Mon Nov  2 21:03:11 2015

Lastly, my original fault regarding arithmetic still works :-

oregano.shadymint.com.combo=\
(\
  oregano.shadymint.com.http +\
  oregano.shadymint.com.bbd +\
  oregano.shadymint.com.procs\
)\

=3
bash-3.00$ combostatus --no-update --debug

23780 2015-11-02 21:13:20.875867 Opening file /apps/xymon/server/etc/combo.cfg
23780 2015-11-02 21:13:20.881199 Symbolic '(oregano.shadymint.com.http+oregano.shadymint.com.bbd+oregano.shadymint.com.procs)>=3' converted to '(1+1+1)>=3'
23780 2015-11-02 21:13:20.881490 Adding to combo msg: status oregano,shadymint,com.combo green Mon Nov  2 21:13:20 2015

-- 
Andy

On 11/2/2015 11:23 AM, J.C. Cleaver wrote:
- snip -

I'm building the RC2 right now. It tumbled out the following warnings:

loadalerts.c: In function 'criteriamatch':
loadalerts.c:822:9: warning: assignment discards 'const' qualifier from pointer target type
  dgname = textornull(xmh_item(hinfo, XMH_DGNAME));

xymonping.c: In function 'show_results':
xymonping.c:363:5: warning: format '%u' expects argument of type 'unsigned int', but argument 2 has type 'long unsigned int' [-Wformat=]
     printf(" (%u usec)\n", rtt_usecs);
     ^

I'm continuing on my way, but thought I'd let'cha'know.

-- 
    Do things because you should, not just because you can.

John Thurston    XXX-XXX-XXXX
user-ce4d79d99bab@xymon.invalid
Enterprise Technology Services
Department of Administration
State of Alaska

On 11/2/2015 11:23 AM, J.C. Cleaver wrote:

My testing with 4.3.22-RC2 and the patch shows expected parenthesis 
behavior is back. Lovely :)  I also continue to see correct arithmetic 
behavior in combo.cfg

I am testing out the much awaited (by me, anyway) noflap flag.

-- 
    Do things because you should, not just because you can.

John Thurston    XXX-XXX-XXXX
user-ce4d79d99bab@xymon.invalid
Enterprise Technology Services
Department of Administration
State of Alaska

On 10/29/2015 5:44 PM, J.C. Cleaver wrote:

Changelog is at
https://sourceforge.net/p/xymon/code/7707/tree/branches/4.x-master/Changes

I suspect the man-page for xymongen has a minor typo for --no-showemptygroups

Where it says "When groups are hosts are made", I suspect you mean to say, "When groups of hosts are made"

It's a nice option and I don't object to its behavior, but since it reverses the historic behavior of xymongen, I wish it would get a little more prominence in the changelog.

-- 
    Do things because you should, not just because you can.

John Thurston    XXX-XXX-XXXX
user-ce4d79d99bab@xymon.invalid
Enterprise Technology Services
Department of Administration
State of Alaska

On Mon, November 2, 2015 2:42 pm, John Thurston wrote:

Good to hear! I'll add it in.

-jc

On 10/29/2015 5:44 PM, J.C. Cleaver wrote:

I have another 24 behind me on 4.3.22-RC2 on Solaris. It is interacting well with my 4.3.21 (Solaris) and 4.3.17 (CentOS) instances.

I've updated some of my production http tests to handle the change in behavior for 301 responses. The no-showemptygroups option was needed on my xymongen to regain the previous behavior. I've deleted some new .sh scripts from cgi-secure. And I'm loving the noflap tag.

Has anyone else anyone turned up oddities with 4.3.22-RC2 ?
Has anyone else deployed it?

-- 
    Do things because you should, not just because you can.

John Thurston    XXX-XXX-XXXX
user-ce4d79d99bab@xymon.invalid
Enterprise Technology Services
Department of Administration
State of Alaska

Hi,

On Tue, Nov 03, 2015 at 03:36:17PM -0900, John Thurston wrote:

Has anyone else anyone turned up oddities with 4.3.22-RC2 ?

Not yet.

The new view on which http status code is bad and which one is good is
a little bit annoying but it actually dug up an unexpected 403
Forbidden on some directory listing. :-)

Has anyone else deployed it?

I have 4.3.22-beta on a few client hosts for a few days now,
4.3.22-rc2 there since yesterday evening. 4.3.22-rc2 on two production
servers since about 30 minutes ago. Looks fine so far. Still fighting
with the httpstatus codes there, though.

		Kind regards, Axel Beckert
-- 
Axel Beckert <user-96d9963fe797@xymon.invalid>       support: +41 44 633 26 68
IT Services Group, HPT H 6                  voice: +41 44 633 41 89
Departement of Physics, ETH Zurich
CH-8093 Zurich, Switzerland		   http://nic.phys.ethz.ch/

Now that Xymon 4.3.24 is out, I'm still seeing the same compile warning 
previously noted in this thread:

xymonping.c: In function ‘show_results’:
xymonping.c:363:5: warning: format ‘%u’ expects argument of type 
‘unsigned int’, but argument 2 has type ‘long unsigned int’ [-Wformat=]
      printf(" (%u usec)\n", rtt_usecs);

I'm not sure what the impact is.

Thanks,  Tim

On Sat, December 5, 2015 12:31 am, tgm wrote:

xymonping.c: In function ‘show_results’:
xymonping.c:363:5: warning: format ‘%u’ expects argument of type
‘unsigned int’, but argument 2 has type ‘long unsigned int’

Thanks. There shouldn't be any real impact for you.

Now that fping is stable and being maintained again, xymonping is edging
more towards quasi-deprecation unless there's a reason it's needed on
specific platforms. This should still be fixed, though. Thanks for
pointing it out.


Regards,
-jc