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

[asmoranomar]

The wierd thing is that image files in a web page aren't required to include the full URL, the browser knows that it comes from the current page. This actually works fine in my IE and Firefox. More than likely is working fine in yours in most situations.

All the pictures (or php file producing the pic) in my cacti installation are being called by the full URL.

The picture (well php file generating the pic) is already in the current working directory. Correct me if i'm wrong but wouldn't it either be the full URL or just the calling the current file work? If /cacti/ is already the current working direcotory, you specify the directory again, would it not make the server try to pull the file from 'http://server/cacti/cacti/blah.php'? (altho i have to admit, you'd see that in the html source, but I didn't get that either)

I'm using IE (6.0) at the moment, but the apache server is what is locking up. I can test this by pulling the page on one PC and attempting to load a plain old test HTML file on another PC and the server wont respond or indicate a connection attempt. I can also let either PC 'sit' and finally time out, and navigate other websites just fine, but I can never get back to my server after (so it's not a browser crash). I'll try to use firefox, but (for reasons I cannot explain) it is not allowed on our network and switching is not an option for us.

I hate to be the squeeky wheel but this plugin is prolly the best feature to ever be added to cacti. Thank you for the hard work you put into it.

[asmoranomar]

I just used Firefox 1.5 and the server still locked up. I tried this on two machines, in the same manner as IE. One 'locked' up and the other could not get a simple HTML file to load. Both could browse the web after it timed out.

Interesting note however - Firefox's image property indicates the URL is correct, but it wont let me view the page's source at all (meh). I dont trust it without the source however. [edit - I finally got the source after lots of trial and error, the Firefox and IE source matches. Again, there is no http://server/ in the part with the picture. Why it says two different things in the properties and the source i have no clue. Source from any other page in cacti has the full URL for images]

Also - I think I mentioned this before - but apache logs dont show IE or Firefox trying to pull this picture file. So maybe it occurs before then?

[edit] I forgot to ask - does it really hurt if I add the full URL to the cacti directory vs just the directory? It doesn't 'look' like it hurts but you would know a little better than I would. If it doesn't then I've already found my 'fix'.

[torstentfk]

Hello (cigamit?)

if I use the new thold plugin I get no host down notifications. I changed to some version back and it works again. Could you please check the changes between the last three versions?
Torsten

What 2 versions? That is what will tell me what I changed. Ofcourse the host down notifications is about to be removed from the threshold plugin, and added to the monitor plugin, since that is what it does primarily, monitor down hosts.

The latest version does not work on my site. I think v. .2.6 works fine.

[fBSDmon]

Hi all,

I've been using Cacti 0.8.6h (Plugin Architecture - Monitor, Reports, Thresholds) on FreeBSD 5.4 for less then a week now and I'm amazed from what I've seen. As I'm thinking of fully switching to cacti (using mostly custom made solutions). I would like make some suggestions on what I think will be some great futures in cacti.

I would like to see some kind of user profiles system for thold and monitor.
Example: The admin/user can choose at what time intervals, which alerts (host up/down, tholds) to receive and where to receive them (e-mail, sms, syslog/eventlog, page). A solution like this can be found in MetaNAV, which is too complex for my taste.

Also, one question:
What is the easiest/fastest way to make thold and monitor alerts go in a table?

Why: I have a perl script that polls (every N sec.) new messages from a table, runs them trough N regexp filters (determines which filters the message belongs to), and then sends e-mail, sms or both to the users that have been subscribed to those filters, messaging methods.
Simple, but it works

I thought of using syslog, but the monitor plugin doesn't seem to send host up/down events. All I'm left with is an e-mail proxy, which I'm not so fond of because it will add more delay to my alerts.
So, before I start poking around the code, I would like your suggestions or a hack maybe

Thanks!

[torstentfk]

Hi Big Developer,

some versions ago thold mailed the alarms with embedded pict in the way that outlook displays the pic at the preview.
Now the pictures are attached and a html-file is attached to. One must open the pict per mouseclick and could not be viewed in preview mode.
Could this feature be reenabled to use the preview-mode?

Torsten.

PS: The spam-check could not be verified on our side because the mail gw has learn that thold ist not a spam!

[cigamit]

Hi Big Developer,

some versions ago thold mailed the alarms with embedded pict in the way that outlook displays the pic at the preview.
Now the pictures are attached and a html-file is attached to. One must open the pict per mouseclick and could not be viewed in preview mode.
Could this feature be reenabled to use the preview-mode?

Torsten.

PS: The spam-check could not be verified on our side because the mail gw has learn that thold ist not a spam!

I have't changed the way the emails are formatted in about 6 versions, it does send out as a multipart, which includes both a html and a text version. The HTML version still has the graph embedded in the email. Is your outlook now configured to only show the text mode?

[torstentfk]

Hi Big Developer,

some versions ago thold mailed the alarms with embedded pict in the way that outlook displays the pic at the preview.
Now the pictures are attached and a html-file is attached to. One must open the pict per mouseclick and could not be viewed in preview mode.
Could this feature be reenabled to use the preview-mode?

Torsten.

PS: The spam-check could not be verified on our side because the mail gw has learn that thold ist not a spam!

I have't changed the way the emails are formatted in about 6 versions, it does send out as a multipart, which includes both a html and a text version. The HTML version still has the graph embedded in the email. Is your outlook now configured to only show the text mode?

No i can view html mails automaticly. I did not change something on outlook. Take a look at the pict...

Torsten

[cigamit]

Hi Big Developer,

some versions ago thold mailed the alarms with embedded pict in the way that outlook displays the pic at the preview.
Now the pictures are attached and a html-file is attached to. One must open the pict per mouseclick and could not be viewed in preview mode.
Could this feature be reenabled to use the preview-mode?

Torsten.

PS: The spam-check could not be verified on our side because the mail gw has learn that thold ist not a spam!

I have't changed the way the emails are formatted in about 6 versions, it does send out as a multipart, which includes both a html and a text version. The HTML version still has the graph embedded in the email. Is your outlook now configured to only show the text mode?

No i can view html mails automaticly. I did not change something on outlook. Take a look at the pict...

Torsten

The only question I have is why the bottom picture is showing in "(Rich Text)" instead of "(HTML)". Do you think its possible your Spam filter is converting them?

[torstentfk]

it could be possible. I have no access on the spam filter
it seems that the filter? put the thold mail into a new mail and send it to the org-email receivers.

[TFC]

Hi,
I can get thold alarms via e-mail like this :

M_2D_7507 - Traffic - 10.200.99.9 - PO6/0/0 went above threshold of 191250 with 1659444

Truly, I set high thold alarm as 191250.
My ask is, how can we define or change these values (191250 and 1659444) as human readeble in mail topic?
Is there any way to do?
P.S. I use thold 2.7

[mrh]

Hi there!
first of all I've got to say: thanks for thold plugin, it's cool!

I'd appreciate very much if someone knows how to make gnokii get alerts generated by thold and send them through SMS. Maybe any kind of php scripting could make this possible?

Thanks in advance!

[clicker666]

I'm having a nasty little problem with my install. I'm running the H version of Cacti, with the latest patches. I'm running my Cacti on Windows XP Pro SP2 with IIS 5.1. I have no problems with Cacti, easily monitoring a number of Canopy devices, until I try to install Threshold.

I run through the installation, line by line, without any problems. I can enter into Cacti and view all the new tabs. Once I click on Save under the Settings tab for Threshold, whether or not I actually save anything, I get the following error: PHP has encountered an Access Violation at 015582DF.

At this point IIS is hung, and I have to restart a number of services and kill processes before I can free it up. If I attempt to connect to Cacti following the service restarts I get the same error. I'm basically forced to restore to my pre-Threshold installation and db's to return to normal.

Any ideas?

[clicker666]

I managed to get it working using the 2.7 plugin. Now I just have to figure out why I'm getting all these false positives.

[rvisintine]

I noticed in the source that you have Graph ID hidden away. I went in and made it visible again as it's very valuable to a perl script I have that parses the Thold page for up/down status and current value of various nodes. The parse module I am using ignores HTML tags so I have been unable to get the link to the graph. Now that I have the graph ID I can just plug it into a generic URL path and have my graph link again.

I guess the point of my ramblings is... don't take it out. I'm sure others would be interested in it too if they knew it was there.

Thanks for the hard work on Thold, it's been a great help to me.

[DanielG]

I was wondering what was the design choice for having the tresholds only in rounded integer.

I'm working on errors on telecom links, where 10 errors by 300 seconds are important values.
For now I've made an ugly patch in the code to multiply value readed from rrd by 301.(':oops:')
It fullfill this niche, but thats not very confortable.

What about to have all the thold plugin working in floating point. Any idea about it?