create kernel panic on ESXi

There are situations when you need to check cluster reactions after a ESX host crash. For example to see if HA will start VM on other hosts.

The easiest method is to pull a hosts powercord. But there ar more elegant ways to let a host crash.

Warning! Do not use on productive systems! This is for testing purposes under controlled conditions only. Use at your own risk.

PSOD

You can trigger a Purple-Screen-of-Death (PSOD) by issuing a special command that causes a kernel panic.  Use the VMkernel Sys Info Shell (vsish).

First you need a SSH connection to your host. Change to vsish

vsish
set /reliability/crashMe/Panic

Alternatively you can issue the command together with parameters.

vsish -e set /reliability/crashMe/Panic 1

Your host will end up in a PSOD and can be restarted afterwards.

 

Backup vCenter Server High Availability nodes

Using Veeam Backup to protect VCHA

The vCenterServer Appliance (VCSA) is becoming more and more important. Many vSphere-related services rely deeply on the availability of vCenter. So it was a logic move to make vCenter more available by having an active, a passive and a witness node of vCenter in your cluster. In case the active appliance breaks, the passive appliance will take over.

I’ve been experimenting with VCHA for a while. Although I’m not yet satisfied with the failover time (5-10 minutes), I think it is a move into the right direction. I wouldn’t call it “high” availability right now – maybe “elevated” availability. But nevertheless the failover time might be alright for many environments today and it is used for production. That raises an important question:

How to backup a vCenter High Availability node?

If you have a standalone VC you’ll just make a backup or replica and you’re finished. But VCHA consists of 3 components and before backing up you’ll need to find out which one is the active node, because VMware only supports VADP-based backups (VMware vStorage API for Data Protection) of the active node. There’s no point in backing up the passive node or the witness. Keep in mind that active and passive node may have switched roles since last backup! Continue reading “Backup vCenter Server High Availability nodes”

Query NTFS Blocksize

Using 64k blocksize for better performance

Partitions with Veeam-Backups, SQL-Database files or SQL-Logs should be formatted with a 64k blocksize for better performance. You can check blocksizes with a simple command.

Open a CMD Shell with Administrator rights.

fsutil fsinfo ntfsinfo <Drive>

Example for the system partition.

fsutil fsinfo ntfsinfo C:
NTFS-Volumeseriennummer : 0xa892e42c92e400a4
Version : 3.1
Anzahl der Sektoren : 0x000000001dc807ff
Gesamtzahl Cluster : 0x0000000003b900ff
Freie Cluster : 0x0000000000b42c71
Insgesamt reserviert : 0x0000000000000ff0
Bytes pro Sektor : 512
Bytes pro physischen Sektor : 512
Bytes pro Cluster : 4096
Bytes pro Dateidatensatzsegment : 1024
Cluster pro Dateidatensatzsegment : 0
MFT-gültige Datenlänge : 0x0000000017180000
MFT-Start-LCN : 0x00000000000c0000
MFT2-Start-LCN : 0x0000000000000002
MFT-Zonenstart : 0x0000000000eb7240
MFT-Zonenende : 0x0000000000eb9340
RM-Bezeichner: FFFF0956-A102-11E7-87BD-005056C00008

You can see Bytes per Cluster: 4096 which equals 4KB.

Script

To check all local partitions for the corresponding blocksize you can use a PS script, which was originally posted by Stuart Moore.

$wql = "SELECT Label, Blocksize, Name FROM Win32_Volume WHERE FileSystem='NTFS'"
Get-WmiObject -Query $wql -ComputerName '.' | Select-Object Label, Blocksize, Name

Links

Stuart Moore – Get Cluster size for all disks and volumes on a Windows machine using powershell and wmi

Veeam Backup & Replication Best Practices