Thold 0.6.1 beta4

computer_guru · Post by **computer_guru** » Fri Sep 11, 2015 7:14 pm

--- 0.6.1 --- [proposed version]
feature: delayed host down alerts - so you don't get host down alerts for problems that quickly resolve themselves. Set the delay per notification list.
beta 2: cleanup, and better debug messages
beta 3: resolved a major bug where delayed down alerts behaved properly, but host up alerts were sent even if a delayed down alert was unsent (still in queue).
beta 4: hopefully I have resolved the bug where host up messages went to the wrong notification lists. I am now initializing all variables.

Thank you cigamit for creating the thold plugin. Hopefully my contribution will help people!

Please test and let me know how it goes!

cigamit · Post by **cigamit** » Fri Sep 11, 2015 7:48 pm

First, please don't just make up a version number, you will confuse others further down the road.

2nd, why would you need delayed alerts? That is exactly what the "Breach Duration" setting is for in each threshold. You make it so each threshold would need to be in breach status for even longer before an alert is actually sent.

computer_guru · Post by **computer_guru** » Fri Sep 11, 2015 10:37 pm

sorry about the version thing.

Delayed threshold works great already, but I wanted delayed host down messages. I've wanted it for a long time. My mistake for not being more clear.

What version number should be used?

I could try to make a new plugin if you prefer, but I really like the notification lists in thold.

cigamit · Post by **cigamit** » Fri Sep 11, 2015 10:45 pm

Ah, delayed Host Down. Ya, normally you would change the Cacti settings for detecting a host down, but that has been problematic for some people. I will look over your changes some more, and then roll them into the current Git Repo. Its about time we have another release as it is.

Thanks!

computer_guru · Post by **computer_guru** » Fri Sep 11, 2015 10:52 pm

Sorry for going about it the wrong way.

That would be awesome if you would go through my changes. I'm sure there is a bug or two. I did say "beta".

. I have some extra logging that probably needs removed or wrapped in a debug block.

Thanks for all the Cacti support you do.

I will test it more on Monday.

computer_guru · Post by **computer_guru** » Mon Sep 14, 2015 11:26 am

I found the function thold_debug so I'm cleaning up my code.

will have the changes done soon.

computer_guru · Post by **computer_guru** » Mon Sep 14, 2015 12:11 pm

Ok, I've updated the attachment at the top of this thread with beta2. much more polished version. better debug messages that only appear when Thold's debug setting is enabled.

also, delay selection box now goes from 5 minutes to 3 hours.

computer_guru · Post by **computer_guru** » Mon Sep 14, 2015 4:38 pm

cigamit,

What do you think about adding a feature where each notification list can be designated as HTML or Plain Text or Brief (SMS) ?

computer_guru · Post by **computer_guru** » Tue Sep 15, 2015 3:57 pm

beta3 - uploaded. see top post for details.

To anyone trying to beta test this: let me know if you run into any issues.

computer_guru · Post by **computer_guru** » Thu Sep 17, 2015 1:48 pm

I'm investigating a possible bug.

I just switched to a new Cacti server. During the transition all hosts showed as "down" because the server NIC was not setup for a few minutes.

When all 400 hosts came backup, the "returned from down" messages did not honor the correct notification lists. Many of the message went to too many recipients / notification lists.

I'm not sure if I introduced this bug or if the bug was already present in a situation where there are so many emails being processed.

computer_guru · Post by **computer_guru** » Thu Sep 17, 2015 7:10 pm

beta4 release - fixed bug that was present if a large number of up/down notifications went out at the same time (my cacti server was offline and marked all 400 hosts as down), then the notification lists were scrambled/wrong.

It appeared to be appending:

wrong behavior: 1st email = list1; 2nd email = list1, list2; 3rd email = list1, list2, list3 (i.e. it would use the correct notification list + all previously used lists)

correct behavior with bug fix: 1st email = list1; 2nd email = list2; 3rd email = list3 (or whatever list it should have used)

computer_guru · Post by **computer_guru** » Tue Dec 01, 2015 5:54 pm

The latest version (beta4) appears to be working well for me. I have not heard feedback from others yet. I hope it is working well for others as well.

ikorzha758 · Post by **ikorzha758** » Wed Dec 02, 2015 1:30 pm

Computer_guru. It does work well I been using your Thold 0.6.1 beta4 in my prod for the last month all works well. I'm hoping to contribute to the thold plugin project by providing a fixed up version of Thold 0.6.1 beta4 that includes a fix for thold daemon that is currently non-functional in a near future...

tsobolev · Post by **tsobolev** » Mon Dec 21, 2015 7:57 am

On cacti 0.8.8f and beta4 have error 12/21/2015 03:52:09 PM - CMDPHP: Poller[0] ERROR: SQL Save Command Failed for Table 'thold_data'. Error was 'Column 'time_fail_trigger' cannot be null' .
Thold has not been created.

johannesschreim · Post by **johannesschreim** » Mon Dec 21, 2015 2:45 pm

tsobolev wrote:On cacti 0.8.8f and beta4 have error 12/21/2015 03:52:09 PM - CMDPHP: Poller[0] ERROR: SQL Save Command Failed for Table 'thold_data'. Error was 'Column 'time_fail_trigger' cannot be null' .
Thold has not been created.

I get the same error : (

Can someone help us?

Is there anyone who has cacti 0.8.8f + thold 0.5 running??

Cacti

Thold 0.6.1 beta4

Thold 0.6.1 beta4

Re: Thold 0.6.1 beta1

Re: Thold 0.6.1 beta1

Re: Thold 0.6.1 beta1

Re: Thold 0.6.1 beta1

Re: Thold 0.6.1 beta1

Re: Thold 0.6.1 beta2

Re: Thold 0.6.1 beta2

Re: Thold 0.6.1 beta3

Re: Thold 0.6.1 beta3

Re: Thold 0.6.1 beta4

Re: Thold 0.6.1 beta4

Re: Thold 0.6.1 beta4

Re: Thold 0.6.1 beta4

Re: Thold 0.6.1 beta4

Who is online