cheat_sheet.txt

Opening slide
=============

- What is a VM crash?
- It is really a catastrophic failure of the VM - no way to recover.
- The best we can do is to collect as much data as possible in order to analyze/reproduce the problem
- Notice that this is different from a Java program error like an unhandled excetion.
- The latte will still shut down the VM, but it will not compromise its integrity
- If you adhere to the Java standard (i.e. no Unsafe :) the VM should never crash (JNI is an exception here of course )


Crash
=====

- Ever seen this: you run a simple Java program and suddenly BOOM
- Who of you has ever seen this in his logs (or found hs_err files in the file system)?
 
- So this is the crash summary you see on the console/log file
- It tells you:
  - what happened
  - where it happened
  - is there a core file and if yes where is it
  - where's the full crash report
    - this can be tricky (look at /proc/sys/kernel/core_pattern)
    - check ulimit


How are hs_err files generated
==============================

- if the HotSpot detects an inconsistent state or its integrity is compromised
  - because of unexpected signals (keep in mind that HotSpot heavily uses signals itself!)
  - because of failed assertions/guarantees (effectively assertions in the product build)
  - because there's not enough native memory

- use SafeFetch

Native (mixed) Stack Trace

Java Stack Trace

Disassembly (hs_dis)