Enhance Notification behaviors per status.

[xellos00]

• Parents: Enhancement on Dispatchers(Notifications. Prev) #240
• Related: Critical or Warning Notification behaviors include reminders.  #163

---

Checklist

• New Feature for the Service/Plugin (Vatz)
• Enhancement (Vatz)
• others(etc. e.g, documentation, project policy management)

---

Is your feature request related to a problem? Please describe.

A clear and concise description of what the problem is. such as Ex. I'm always frustrated

This issue only enhancement for current behaviors that mentioned at #163 (comment)

Describe the solution you'd like| Ex

A clear and concise description of what you want to happen

• AS-IS: Notification send alert only following cases

State	Severity	Send Notification	Condition
!SUCCESS	CIRICAL	Y
!SUCCESS	ERROR	Y
SUCCESS	---		anything that send recover

---

• TO-BE:

State cases (We are going to use only following flags until there's cron job or plugins has process that rely on 3rd application)

• SUCCCESS
  • INFO : Only if everything is okay, getting metric and in success status.
• FAIL
  • CRITICAL: fail to get appropriate metric value and it's critical
  • WARNING : fail to get appropriate metric value and it's warning
• ERROR
  • CRITICAL : gets error to get metric info from plugins.

State	Severity	Send Notification	Condition
SUCCESS	INFO	Y	anything that has recovered from Alert (1 time)
FAIL	CRITICAL	Y	Fail to get appropriate metric value and it's critical (e.g 95% of disk usage)
FAIL	WARNING	Y	Fail to get appropriate metric value and it's warning (e.g 70% of disk usage)
ERROR	CRITICAL	Y	Get error to get metric Info (so any error that gets from plugins has to send response with this)

[xellos00]

@dsrvlabs/validator
is there any other suggestion table above?

[gnongs]

@dsrvlabs/validator is there any other suggestion table above?

If 95% of disk usage, Isn't state SUCCESS and severity CRITICAL ?

[rootwarp]

I think meaning of Error and Critical seems little bit confused
because I don't know which one is more serious level.

How about

• Critical or Fail
• Warning,
• Success

[heejin-github]

May I ask about Failed condition of state?
It means failed to get appropriate metric value. but, how do we know that it is Critical or Warning? even if we didn't get the value.
I think that if the state is Failed, we couldn't identifying the severity.

[xellos00]

@Choi-Jinhong @heejin-github @rootwarp
Thanks for patients, I almost forgot to reply for this.
Let's summarize alert behaviors in table.

[xellos00]

enum STATE {
    NONE = 0;
    PENDING = 1;
    IN_PROGRESS = 2;
    SUCCESS = 3;
    FAILURE = 4;
    TIMEOUT = 5;
}

enum SEVERITY {
    UNKNOWN = 0;
    WARNING = 1;
    ERROR = 2;
    CRITICAL = 3;
    INFO = 4;
}

1. flags(State + Severity) 조합을 기준으로 노티를 받는 상태값 정리
2. Y Conditional notification이 있는데, 그것에 대한 명확한 Condition을 정하기 위함.
3. Send recovered message when State & Severity is SUCCESS | INFO and stop reminder.

State	Severity	Send Notification (Y/N)	Recovery (Y/N)
SUCCESS	WARNING	Y	Y
SUCCESS	CRITICAL	Y	Y
SUCCESS	INFO	Y	N
FAILURE	*	Y	Y
ANY OTHER	*	N

---

Condition
Send a single time notification when any flag(state+severity) has changed from previous flag

1. SUCCESS

• SUCCESS | WARNING |
• SUCCESS | CRITICAL |
• SUCCESS | INFO |

2. FAILURE

• FAILURE | * |