Generating a crash dump

Modified: 26 Sep 2019 21:20 UTC

Any time a system becomes nonfunctional and has to be rebooted, you must write a crash dump so that Joyent support can diagnose the issue. Without a crash dump, diagnosing system hangs is effectively impossible.

You can write a crash dump using the console, or by generating a NMI (non-maskable interrupt).

Generating a NMI using the console

If you can log in to the system console of the unresponsive node, you can cause a system dump.

To do so, log in to the system console of the unresponsive compute node and enter one the following commands:

# halt -d

or

# uadmin 5 1

Generating a NMI via IPMI

Joyent systems are configured to generate a crash dump when they receive a NMI. To issue an NMI through IPMI, run the following command from a remote system:

ipmitool -H ${SERIAL_OVER_LAN_DEVICE_IP} -U root -I lanplus chassis power diag

Generating a NMI via IPMI

Some systems have a hardware button that you can use to generate an NMI. Please check your hardware vendor's documentation to determine if your hardware has this feature for your particular server model.

Checking the location of a crash dump

savecore stores crash dumps extracted from the dump device in /var/crash/volatile.

You can use dumpadm to check the location of a crash dump. Enter one of the following commands:

$ pfexec dumpadm
      Dump content: kernel pages
       Dump device: /dev/zvol/dsk/zones/dump (dedicated)
Savecore directory: /var/crash/volatile
  Savecore enabled: yes
   Save compressed: on

or

$ cat /etc/dumpadm.conf
#
# dumpadm.conf
#
# Configuration parameters for system crash dump.
# Do NOT edit this file by hand -- use dumpadm(1m) instead.
#
DUMPADM_DEVICE=/dev/zvol/dsk/zones/dump
DUMPADM_SAVDIR=/var/crash/volatile
DUMPADM_CONTENT=kernel
DUMPADM_ENABLE=yes
DUMPADM_CSAVE=on
$