Information Central - notifications
-
As discussed at other places, I open here a topic, which @nebulon asked me about to do, to find out what kind of Information the admins wish to get informed about and in which time frame.
Based on my experience which was described here and what @msbt described in the same discussion,
I came back from a short trip and thought everything would be okay (since I didn't get any emails that said otherwise) …
I put a few working thesis:
- Information is the road to (pro-)actively manage your fleet and deliver services without interruption/minimized down-times.
- Information hast to be delivered to the admin in an organized way, so that he/she can react a.s.a.p., but should not be flooded with not system critical info.
- Every change to a running system is essential to know about.
- Every failed task needs to be reported immediately.
To handle different cases where an admin want every serious incident reported at once, and some admins could change, what they want to get informed about, it maybe would be a forward-locking investment in development, to have a central point of management for the notifications.
as a starting point: organise notifications:
- have different severity classes
- have different channels like
- different email-addresses
- forward to messengers, telegram or matrix, because they are read/recognized faster
- configurable classes/channels relationship
- configurable triggers, like on first incident, or only at the third …
before the CLDRN information-stream was cut down, I used sieve-scripts to mimic something like severity-classes. it's somehow cumbersome to manage and had no other channel, meaning also, that if the host's mail system had a pbl, the flow was doomed.
To picture the other side, I am very happy with what LibreNMS has to offer with their customizable alerting system and use that with a telegram robot to inform me about any system-critical information on my fleet at once.
one can read more about it: customizable alerting system
What solution would fit for cloudron, without overaggregating it, but delivers essential info at once and not so critical info to another channel, to have something like a sysadmin's diary?
-
@chymian said in Information Central - notifications:
forward to messengers, telegram or matrix, because they are read/recognized faster
We discuss this in another post, I have proposed to offer a Webhook interface for events.
And it was on the plan for update 6.3 but then postponed: Gitlab issue