Maintenance window between event START and event END

[AntonPineiro]

Hi,

I would like to know your opinion about some automation use case we have. Hosts are patched during an undefined window:

• It can take minutes, hours or days.
• We do not know start time or ending time.

Our idea is:

• Sending event to Dynatrace when activity starts + activity ends.
• Create a maintenance window during that range to avoid problems are sent out Dynatrace.

Requirements:

• Problems can be created in Dynatrace, in the UI. But not trigger any problem notification outside (Email, ServiceNow, Slack...).
• Maintenance window will cover host partially, not entire host (some proceses, or child entities).

Possible solutions:

1. Create a maintenace window per host using workflows (but we need to clean up maintenane windows time to time).
2. Custom extension ActiveGate for filtering in/out problems raised during patching.
3. Site Realibity Guardian? Can be an option when evaluation time is undefined? Can take minutes, hours or days.

Have you found a similar scenario before? Do you have another alternative?

Thank you!

Best regards

[JR79]

Tag based maintenance could be an option but it doesn't work as it will not re-evaluate open problem when you remove the tag.

There is just no solution for this right now in Dynatrace that could work on several large multi tenants environment.

[rastislav_danis]

I actually use version of this my older solution based on management zone filtered by host tag (tag is not directly mapped to Maintenance Window, which gives you freedom of tag propagation) on customer's managed: https://community.dynatrace.com/t5/Alerting/Suppression-of-Problems-during-monthly-OS-patching/m-p/2... 

[AntonPineiro]

Hi,

Thank you for your feedback. One question about this, what happens in this scenario?

1. Problem is created but no alert because entity has the tag, it means, problem is ignored.
2. After 3 hours tag is removed because patch has finished, but something was wrong, it means, it is a real alert now.
3. Issue is still there but problem was created under maintance window, it means, no notification because it had been created during a maintenance window, and maintenance window still exist. How do you re-valuate problem to get notified? Some "watchdog" workflow?

Best regards

[rastislav_danis]

re-evaluation of maintenance window is not possible afaik. You can only try to narrow patching window to be as small as possible via adding/removing tag into patching procedure, so it will trigger maintenance window only for neccessary patching time.

[JR79]

Re-evaluation of alert in maintenance windows is happening at the end of the maintenance window but in case the mw is infinite and based on tag, there is no end and therefore no re-evaluation.