Why Are Graceful Reboots Detected as Ungraceful Reboots in Dynatrace

Platform Open Q&A
1

Hello Community,

I observed that even graceful reboot events are being reported under the “Ungraceful Reboot” problem category in Dynatrace.

I verified this behavior across the last 365 days of alerts, and the pattern remains consistent — graceful reboots are still classified as ungraceful.

Could someone please explain:

  • How Dynatrace determines whether a reboot is graceful vs ungraceful?
  • Under what conditions a graceful reboot may still be detected as ungraceful?
  • Are there any known limitations, configurations, or log dependencies that can cause this behavior?

Any supporting documentation or best-practice references would be greatly appreciated, as this information is required for internal and customer validation.

Thank you in advance for your support.

2

Hi @Sohel_Rashid,

Some answers 😉

How Dynatrace decides “graceful” vs “ungraceful”

  • Graceful: the OS sends a shutdown signal that reaches OneAgent → OneAgent can emit a graceful shutdown / reboot graceful event.
  • Ungraceful: the previous shutdown was unexpected (no shutdown signal received), e.g., crash/power loss → reboot ungraceful.
  • Classification is based on host availability events and system-specific logs/events used to detect reboot type (Linux/AIX/Windows supported).

Why a “graceful” reboot can still appear as “ungraceful”

  • If OneAgent receives no shutdown signal, Dynatrace classifies it as unexpected (→ ungraceful).
  • Linux/K8s timing/network ordering: if the network manager is stopped too quickly (or there are connection problems), the shutdown event may not be sent in time, leading to ungraceful detection after reboot.
  • K8s node termination: OneAgent uses an inhibitor lock to gain time; insufficient rights can break that protection and cause missed/late shutdown signaling.

Known limitations / fixes worth checking

  • Dynatrace has fixed cases that could generate false/incorrect “Host ungracefully rebooted” problems, including:
    • OneAgent installed within 5 minutes of boot → could trigger “Host ungracefully rebooted” (fixed).
    • Problems created for ungraceful reboots when a host is not monitored (fixed).
    • Reboot detection issues due to different OS logs flow (fixed).

Configuration note (often confused with classification)

  • By default, Dynatrace alerts on unexpected outages; you can opt-in to alert on graceful shutdowns. This affects notifications, not the underlying “signal received vs not received” classification rule.

I hope it helps you.

Regards,