Saturday, October 24, 2020

Why CoreDump files are useful?

 In a case of any unrecoverable system error, you may see such a blue screen (Windows OS) or a purple screen (Linux OS or ESXi) that shows some useful information about what's going on the server and why this crash happened? In most failures, there is a hardware reason for this problem, and many of them are related to the memory and storage devices or even processors. So there is a possibility to register this diagnostic info to a Core Dump file if you configure Disk Dump or Net Dump in the ESXi host.

In addition to the mentioned reasons, many other problems may exist for the crashing of an ESXi host, like an incompatibility between the physical server and version of the ESXi host. In most cases of PSOD (Purple Screen of Death), the ESXi host doesn’t reboot and will hang on the error screen until you reboot it manually, while you can see some cases of the automatic restart. However if you couldn't see the error message (while the server is freezed) or didn't have physical access to the host in the server room (via any physical media like monitor or KVM console) then you can use an OOB management platform used by your server's vendor (like Intel AMT, Dell iDRAC and HP iLO) to analyze and check the statistical or current status of the ESXi host.

In older versions, there was a size limitation for default CoreDump (100MB) and as you understand it’s not enough for this important file. So VMware highly recommended to generate them in one of your VMFS datastores. VMware announced "After release ESXi 7.0 it creates a VMFS-L based ESX-OSData volume and configures a CoreDump file to stored in it if the volume is larger than 4GB."
So if the ESXi is installed in a USB device or SD memory card, then setting of the following boot option is required before the host startup:  

allowCoreDumpOnUsb=TRUE  

Also if you need to find and investigate the current CoreDump partition, run the following esxcli command: 

esxcli system coredump partition set 

If you need more details about how to modify CoreDump files, you can read this post too. As the last point, you should consider both Software-based adapters iSCSI and FCoE are not supported for registering CoreDump file until the current ESXi version.


 

No comments:

Post a Comment

I will start a new journey soon ...