[RELEASE] thold-0.4.4-43

[noname]

One solution for "Threshold(s) Already Exist - No Thresholds Created" error that I've faced:
- http://forums.cacti.net/viewtopic.php?p=215196#p215196

It determines whether if threshold is created successfully by validating "strlen($message)".
But if some error occurred while processing, then $message will not be set properly.

[sodium]

for all followers...

just released a small fix for the 0.4.4-43 code

- 0.4.4-45: fix template selection view mechanism

when you were viewing the current triggered thresholds the selection mechanism for the template did something wrong (template was not correctly applied to the data)

have fun

[TheWitness]

Sodium,

Please compare your fork to branches/stable and prepare an update and I will merge. I want to reduce the level of 'forking' of Thold going on. I will incorporate as required into 0.4.6.

Thanks,

TheWitness

[sodium]

Please compare your fork to branches/stable and prepare an update and I will merge. I want to reduce the level of 'forking' of Thold going on. I will incorporate as required into 0.4.6.

we are merging our changes to your SVN tree...

this will take some time to stabilize the code again (you have done some nice improvements in the original code : ) )

hope this is soon enough

[TheWitness]

Sodium,

Please also PM me a dump of your thold structure and copy of latest plugin. Dump command is "mysqldump -d cacti > file_for_witness.sql" which only dumps structure and not data.

TheWitness

[sodium]

Will I do (your command will dump all of the structure of cacti and not the thold section alone)

I'll PM you both

[brunoromero86]

what you describe is almost correct
we determine the rrd steps (normally 5min, 30min, 2hour, 1day) and offer this list of steps (without the current 5 min) as the reference in the past.
this determination is done by reading the (from the rrd file) average back and store it in the database for the validity of the measurement. (thus @ 00.00 hour the new value is placed in the database -- for 1 day average)
every 5 mins we compare the current value with the average stored and determine the % deviation
and this is done to minimize the cpu power needed to check the values

have fun

Hi sodium
I would like to know if you can help me with configuration of THOLD parameters.
I have to monitor the average of the last day and compare with the average of the current day.
The problem i´m trying to resolve is shown on the figure bellow.

I want to detect this kind of problem.

Graphic.PNG (39.03 KiB) Viewed 2461 times

Is there some configuration in current version (Thold 0.5) that I can use to monitor and detect this problem?

Thanks in advance
Bruno

[sodium]

that should work comparing the current value with the 'day avaerage' but IMHO bandwidth graphs (with nightly lows) tend to create false positives
you could tweak the false positives with the "Baseline Trigger Count"

[brunoromero86]

OK Sodium, I´ll try it.
Thanks again!