Thold 2.x - Thresholding/Alerting module for cacti 8.6

[rpingar]

has someone tested Thold with the new cacti 0.86d?

Any comment?

thanks

[t0mm]

.. we remplace argus+bb with cacti + thold.

but, what about escalading !?

i'v modified a little bit you'r last check_thold.php to permit
me using thold_fail_count & thold_fail_ack (new field) to escalade.

do you have any plan in this way ?

BTW, thanks a lot, with small code you make big smile here

[t0mm]

i forget to say that,

We poll about 3800 sources in ~11s ( i'm also graphing the time to poll heh

[JimmyC]

I installed Thold on Cacti 0.8.6d.
So far no problems. everything is working.

[TheWitness]

Cacti 0.8.6d is a maintenance release. There are primcipally only bug fixes, spike prevention (now with cactid, look under announcements) and rrdtool 1.2.x support. Therefore, THold should continue to work fine.

TheWitness

[Rossco]

Hi guys

I have had THOLD installed and working recently, but I only used it on one rrd as a test and then left it for a while. I've now come to add a number of threshholds and after adding 4 everytime I now click on a template hyperlink to create a new threshhold the whole session hangs(requires closing the browser).

I've deleted all the thresholds I created and tried again, but I cant create a single threshhold now as the browser crashes. Nte: I can click on the threshholds hyperlink, and that displays fine, but I cant reach the section to create threshholds.

In my php error log I seem to be getting this reported

[06-May-2005 07:50:40] PHP Notice: Undefined offset: 0 in d:\inetpub\wwwroot\cacti\thold.php on line 46

It seems bizarre, I've no idea how or why it should suddenly start crashing.

Any suggestions welcome.

Kind Regards

Rossco

[gandalf]

....

[root@cacti-test cacti-cactid-0.8.6c]# make install
source='util.c' object='util.o' libtool=no \
depfile='.deps/util.Po' tmpdepfile='.deps/util.TPo' \
depmode=gcc3 /bin/sh ./config/depcomp \
gcc -DHAVE_CONFIG_H -I. -I. -I./config -I/usr/local/include/net-snmp -I/usr/local/include/net-snmp/.. -I/usr/local/mysql/include/mysql -g -O2 -c `test -f 'util.c' || echo './'`util.c
util.c error: conflicting types for 'mail'
util.c error: previous implicit declaration of 'mail' was here
util.c2: warning: no newline at end of file
make: *** [util.o] Error 1

Got a similar problem. Solved it by moving the code for the "maill" procedure in util.c from the bottom of the file just before the first procedure "read_config_options".
I'm not that good in C, but it seems, that "mail" was used before it was defined.

HTH
Reinhard

[gandalf]

In my php error log I seem to be getting this reported

[06-May-2005 07:50:40] PHP Notice: Undefined offset: 0 in d:\inetpub\wwwroot\cacti\thold.php on line 46

Got the same problem with freshly installed thold 2.1
Using cacti 0.8.6d with cactid and rrdtooll 1.2.1
To my understanding, this seems to be the problem of new rrdtool. Here's an example, first with rrdtool 1.0.49 (working)

Code: Select all

[root@gandalf lib]# /usr/local/rrdtool-1.0.49/bin/rrdtool last /var/www/html/cacti/rra/gandalf_cpu_nice_8.rrd 1115387702
[root@gandalf lib]# /usr/local/rrdtool-1.0.49/bin/rrdtool fetch /var/www/html/cacti/rra/gandalf_cpu_nice_8.rrd AVERAGE -s 1115387702 -e 1115388602
timestamp    cpu_nice

1115387700:  1.7337379107e-01
1115388000:  nan
1115388300:  nan
1115388600:  nan
1115388900:  nan

rrdtool last gets the last update time of rrd. Now the rrdtool fetch gets the data by using last timestamp as a starting point and "last timestamp + 900" as the ending point. In this case, the relevant data ist 1.7337379107e-01.
Now the same procedure with rrdtool 1.2.1:

Code: Select all

[root@gandalf lib]# /usr/local/rrdtool-1.2.1/bin/rrdtool last /var/www/html/cacti/rra/gandalf_cpu_nice_8.rrd 1115387702
[root@gandalf lib]# /usr/local/rrdtool-1.2.1/bin/rrdtool fetch /var/www/html/cacti/rra/gandalf_cpu_nice_8.rrd AVERAGE -s 1115387702 -e 1115388602
                       cpu_nice

1115388000: nan
1115388300: nan
1115388600: nan
1115388900: nan

As you see, no data is returned, because the output starts at the first PDP exceeding the starting point. So as a workaround, I changed line 30 of <path_cacti>/lib/thold-functions.php from

Code: Select all

$last_time_entry = rrd_last($rra, "MAX");

to

Code: Select all

$last_time_entry = rrd_last($rra, "MAX")-3;

Subtracting 3 seconds is just a shot, you may even subtract as much as 300 sec if this is the polling interval.

The difference in rrdtool festch behaviour may as well be relevant for other calls to rrdtool_function_fetch in <path_cacti>/lib/rrd.php.

HTH
Reinhard

[cigamit]

In case anyone is interested, I have done numerous enhancements to our local copy of the Thresholds module, and if you would like to incorporate the code back I'll post them (There are several patches to different files).

1. The Thresholds tab only shows up if you have permissions to access it.

2. Fixed a few problems with pulling graph associations from graph_templates_item when the associations where not inserted into the database in numerical order (a simple "order by id" resolves it)

3. Setup a table for threshold templates, so that we only auto-create thresholds for things that we want. Allows you to set the HI, Low, trigger, notify, ect... (I still need to write the interface to add/modify them)

4. Fixed a few problems with non-associated thresholds. The datasource had been deleted, but the threshold remained, which throws an error, so I fixed the error and it now displays "Unknown Data Source"

5. Modified the Threshold emailing to now use the Open Source class.phpmailer.php (I did modify it all also to add an extra function), which allows me to specify a totally different SMTP Server (with authenication) than the localhost or to just use sendmail (options are good).

6. Using the class.phpmailer.php, now all graphs are emailed as HTML and have the image embedded into the message, so that you see the actual graph, instead of the current (with the old way, they were genereated on the fly, so if you looked at the email 2 days later, you were not seeing the current data). I am currently working on Custom messages with each threshold, since I would rather I was not paged by our NOC if certain thresholds are breached.

7. I have setup a "Monitor" tab (we use it to replace What's Up Gold), which displays all the Hosts you want to monitor (Simple checkbox on the device to disable monitoring). Up devices are green, Down are red, and Recovering are Blue. It throws an Audiable alert when a host is down, you are able to silence the alert, and it uses a little AI to tell when to re-enable the alert automatically. I took the poller, and cmd.php and ripped out alot of stuff to make a "fast" poller which you can run every minute (if you normally only poll every 5). The fast poller only checks to see if the host is up, and updates the database with the info. The fast poller don't interfer with the normal poller as it runs side by side (its not a replacement) The Monitor window will refresh itself every 30 seconds (so to not miss something)

I am sure there are many many other things that I left out, our local copy if heavily modified to make it simple enough for any new employee to add a host, graphs, and thresholds (we also added a tutorials section using Wink). They only have to put in the hostname, ip, and select a host template and click create. All graphs are auto-checked since our data templates and queries are modified to only pull the info we want (only actual HDs, and only actual NICs that are enabled, and the windows services). And now the correct thresholds for the graphs are auto-created with the correct info by a click of a link.

If you would like the patches, I can pull them out of my local CVS and post them. I have a ton of other changes to Cacti itself, it just may take awhile to go through and assemble all the patches.

While I am at it, I would also like to note that I am a developer on the Squirrelmail Team, and a "Plugins Team Leader", so if you are interested in developing a Plugin system for cacti, I can probably help out in that area.

[aboyz]

cigamit

Can you post your patches and stuff and a howto to get this working like yours.. thanks in advance

[ramjack]

Cigamit
Would you mind posting your monitor patch? This is something we would really like to test. Thanks.

[TheWitness]

I must say, makes me tingle!!!

TheWitness

[aboyz]

hi,

anyone got any good working file i can download and try out? if so please post it here so i can try this new plugins out.

thanks

[TheWitness]

The Fetch command written in rrd.php is still working with rrdtool 1.2.x.

TheWitness

[Rossco]

Hi Reinhard

Thanks very much for your comprehensive reply.

Sadly, I have not modified RRDtool at all. I did however try switching between versions and modifying the thold-functions.php you mention, but neither resolves the issue. I have even tried rolling back to a previous SQL database, but that too doesnt work.

I'm now at a loss, the html source on the page only loads so far - it seems to load the left hand menu, and gets most of the way through loading the top, but then stops dead seemingly at the generation of the graph or the creation of the tabs for the data index names. (these only appear on screen when I hit the stop button)

I am on 0.8.6.c on windows 2000 Server, PHP 4.3.9

This is the last section of the html source that loads.

<-----removed---->

<form name="THold" action=thold.php method=post>
<input type='hidden' name='rra' value='11627'>
<input type='hidden' name='hostid' value='309'>
Data Source Description: <br><strong>2600-Reading-Beep - 5 Minute CPU</strong><br><br>
<!-- Threshold: <br><input type='text' name='thold' size='10' maxlength='20' value="" ><br>
-->
Associated Graph (graphs that use this RRD): <br>
<select name='element'>
<option value=11458 selected> 11458 - 2600-Reading-Beep - CPU Usage </option>
</select>
<br>
<br>
</td>
<td>
<img src="graph_image.php?local_graph_id=11458&rra_id=1&graph_start=-32400&graph_height=100&graph_width=300&graph_nolegend=true">
</td>
</tr>
</table>
<table align="center" width="98%" cellpadding=1 cellspacing=0 border=0 bgcolor="#00438C">
<tr>
<td>
<table cellpadding=3 cellspacing=0 border=0 bgcolor="#E1E1E1" width="100%">

I'm about to start going through the whole page and add comments to identify where its hanging.

Many thanks

Ross