Setting up thousands of alerts...
Hi,
I have several thousand subnets I want to monitor for operational time. How is that possible? I was able to create thousands of sites using an external file but I am hearing from Compuware support I have to copy paste into a xml file then hand modify. That just will not work for big companies like mine.
Does anyone know how to do this in a resonable way without taking many hours of hand labor? Seems crazy Compuware wants to do it by hand!
Thanks,
-Tony
Tony,
In Compuware we used to say "manual" and think "everything can be scripted" 
When you were filling the Sites file I bet you were doing some kind of automation ...
Do I get you correctly that all you need is to create one example alert with a name and first subnet and then copy as amny times as many subnets you have making the changes to the IPs and alerts names? If so I beleive this can be scripted i.e. in Bash on Linux.
I understand. I am writing a java program since not all developers use linux.
What I am doing is I created a user defined alert then am going to look at in the custom xml file to see how it is inserted.
Of course the issue is Compuware has a key (alarmid) in there that is auto generated so the person has to hope that the key is the last value and write the script to start from there for an alarm id. I like to have config files maintained by the app seperate and have a bulk loader file like what they did with "sites" function.
I have several thousand subnets so manual entry is a terrible way to go.
No one is perfect but never assume that your product users are developers.
Thanks for the support though.
Hi Tony
I agree with your point that a system that does monitoring of many moving parts should be completely maintainable from the "outside" by using some sort of structured, human readable files.
If you write an ER, I'll vote for it DCRUM Requests for Enhancements
Tony,
If there are parts of the system that may need any kind of "external development" you can always count of Support as we do our best to help in such situations, but I'm happy you're able to manage this situation in Java that I'm personalty don't know so well
On the other hand DCRUM has no bulk way to maintain alarms and as Ulf said it's a great idea to place enhancement request.
Tony,
A huge problem might arise with having such a huge number of unique alerts. I.e. a risk of zdata not being processed by CAS.
It is due to metric alerts being checked during zdata sample processing on CAS.
Thus when having thousands of unique alerts definitions to check it may simply choke or even stop data processing on CAS.
Please also specify what exactly are you trying to achieve.
You might add a Site or Server IP dimension and it will result in alert triggering conditions being monitored for each of the given dimensions.
So you do not need to filter exact IPs or IP ranges.
Regards,
Grzegorz
We have several thousand stores we want to monitor register response times and send an alert if registers slow down. Right now I have entered store subnets that have up to 15 registers on them as sites.I have slightly over 2,000 sites. I plan to setup the alerts based on sites before I saw your response. So as long as the DCRUM product can handle that many sites we should be ok.
Regards,
-Tony
Community Member of the Month
September 2018
Announcing Dynatrace's Community Member of the Month for September 2018, Muthu P.! Click here to read more! Employee Member of the Month
September 2018
Announcing Dynatrace's Employee Member of the Month for September 2018, Kris Z.! Click here to read more! 