Talk given at HashiTalks 2022 by Shai Almog: Learn how continuous observability helps us go beyond the limits of AMS, Logs and other observability tools. Solve real world problems at scale in production in polyglot highly distributed environments such as Nomad, K8S etc. with this free solution.

1. Cloud Native
Debugging in
Production - Dig Deep
into your agents
Shai Almog
January 2022

3. I APMs

7. Developers

8. Developers DevOps

10. Developers DevOps

14. The Needle isn’t
Necessarily Here

15. Production
issues
in 2021 Add new logs,
metrics and traces
1

16. Production
issues
in 2021 Add new logs,
metrics and traces
1

17. Production
issues
in 2021 Add new logs,
metrics and traces
1
CI/CD
2

18. Production
issues
in 2021 Add new logs,
metrics and traces
1
CI/CD
2
Run
application
3

19. Production
issues
in 2021 Add new logs,
metrics and traces
1
CI/CD
2
Run
application
3
Recover
state
4

20. There has to
be a Better Way!

21. We don’t know what
we’ll run into

22. If Only we
Could Debug…

23. But Debuggers aren’t
the Right Tool

24. They Can’t Cross
Server Boundaries

25. They can’t handle
Different Languages

26. I APMs

27. Why Do APMs Impact
Performance?

28. This isn’t Too Bad

30. Developers DevOps

31. Production
Debugging

32. Continuous Observability?

33. Continuous Observability?

34. Continuous Observability?
Continuous Observability is a
streamlined process of constantly
asking new questions and getting
immediate answers.

35. Developer
The developer uses
Lightrun's IDE plugin
to add an action in
example.java line 100
How it Works
1

36. Management
Server sends
request to the
agent
Developer
Management Server
The developer uses
Lightrun's IDE plugin
to add an action in
example.java line 100
How it Works
1
2

37. Management
Server sends
request to the
agent
Agent inserts
the actions
at the specific
location at
runtime
Developer
Management Server
Service Running with
Lightrun’s Agent
The developer uses
Lightrun's IDE plugin
to add an action in
example.java line 100
How it Works
1
2
3

38. Management
Server sends
request to the
agent
Agent inserts
the actions
at the specific
location at
runtime
The data is
transferred to
the developer’s
IDE, through the
Server
Developer
Management Server
Service Running with
Lightrun’s Agent
The developer uses
Lightrun's IDE plugin
to add an action in
example.java line 100
How it Works
1
2
3
4

39. It’s Very Low Overhead?

40. It isn’t an APM

41. What if I Observe
Something Central?

43. Demo: PrimeMain, Kotlin

45. lightrun.com/free

62. Front End JavaScript
with NodeJS

63. The Backend is
Spring Boot

109. Summary

110. Summary

111. Summary

112. Summary

113. Summary
Stability
Inserted Actions are emulated in a
dedicated Sandbox to validate there are
no side effects of the original flow and
state of the process

114. Summary
Stability
Inserted Actions are emulated in a
dedicated Sandbox to validate there are
no side effects of the original flow and
state of the process
Security
Authorization and authentication,
integration with common IDPs

115. Summary
Stability
Inserted Actions are emulated in a
dedicated Sandbox to validate there are
no side effects of the original flow and
state of the process
Security
Authorization and authentication,
integration with common IDPs
Privacy
PII redaction and blacklisting of files /
methods / members

116. Summary
Stability
Inserted Actions are emulated in a
dedicated Sandbox to validate there are
no side effects of the original flow and
state of the process
Security
Authorization and authentication,
integration with common IDPs
Privacy
PII redaction and blacklisting of files /
methods / members
Footprint
CPU footprint is negligible.
Memory and network footprints are
capped and configurable

117. Summary
Stability
Inserted Actions are emulated in a
dedicated Sandbox to validate there are
no side effects of the original flow and
state of the process
Security
Authorization and authentication,
integration with common IDPs
Privacy
PII redaction and blacklisting of files /
methods / members
Footprint
CPU footprint is negligible.
Memory and network footprints are
capped and configurable
Environment agnostic
Operates on-prem / cloud,
microservices, serverless

118. Questions? Shai Almog
Developer Advocate @ Lightrun
twitter.com/debugagent
talktotheduck.dev
github.com/shai-almog
shaia@lightrun.com