Dump restart a virtual machine

Edit online

Starting 2.1.0 release, PowerVC supports restarting a virtual machine by generating core dump file(s). The VM restart with core dump is useful when you want to restart a VM that has crashed. Restarting with dump option creates a kernel dump file on the VM. This file contains the memory, register values, and the call stack of an application at the time of crash.

The VM restart with dump is enabled on all supported guest operating systems.

Prerequisites

HMC version must be V10.1.1020 and higher.

PowerVC version must be 2.1.0 and higher.

NovaLink version 2.1.0 or higher

Before restarting a VM using dump

Before attempting a dump restart, it is recommended to increase the process limits because the default process limits might lead to OutOfMemoryError which leads to truncated core dumps. The truncated core dumps are generally unusable.

Complete these steps to set the ulimit to unlimited.

Run ulimit -c to check if ulimit is enabled.
If the ulimit is 0, run ulimit -c unlimited to enable ulimit.

Configuring dump

This section elaborates details on configuring VM dump on RHEL, SLES, and Ubuntu.

On RHEL and SLES

Check whether the kdump and fadump are properly configured on the LPAR by cloning the following RAS tool that helps with the setup.
```
git clone https://github.com/linux-ras/ServiceReport
```

Install the program.

cd ServiceReport
make build
make install

Once the installation is done, print the status of all plug-ins.
```
servicereport
```
Repair the incorrect configuration after the validation.
```
servicereport -r
```
Note: Complete these additional steps if the guest OS is SLES SP4 or higher.
1. Add crashkernel=2G against GRUB_CMDLINE_LINUX option.
2. Update the grub.
```
grub2-mkconfig -o /boot/grub/grub.cfg
```

On Ubuntu

Install kernel crash dump utility.
```
sudo apt install linux-crashdump
```
During configuration of kexec-tools, select yes to enable kexec-tools for all restarts and enable kdump-tools.
View the status of kdump.
```
kdump-config show
```

The core dump file can be found as given here.

For AIX after dump restart, run sysdumpdev -L for core dump details.
For Ubuntu, SLES, RHEL, the file can be found at /var/crash.
For IBMi guests, the dump manager on IBMi server has a record generated for the virtual machine after dump restart operation is complete.

Restarting a VM with dump

On the PowerVC GUI, navigate to Virtual machines > VM list. Select a virtual machine and then click Restart. On the Restart VM page, select Dump and then click Restart.