Website under construction

Active memory dump 
Another new feature as it relates to diagnostics is the ability to capture memory dumps. Imagine that 
you have a big Hyper-V cluster and each of the nodes has 512 GB of memory. If the node is having an 
issue and you create a memory dump that contains both user and kernel mode memory, that memory 
dump is going to be larger than 512 GB. Trying to work with a file that size can be a nightmare. First, 
you need to ensure that you have a drive with enough free space to hold a file of that size. You then 
need to spend hours zipping, uploading, and unzipping before you can even begin to open it. If the 
problem is with the host server itself and you are running VMs that use 500 GB of that memory, this is 
information that you don’t need, because it does not pertain to the host. 
Because of this, there is a new dump setting called Active Memory Dump. This setting captures only 
the memory that the host is actually using. If the host is actively using only 5 GB of memory, a 5 GB 
memory dump is what will be created. This smaller dump is much easier to parse than the 512 GB file 
in the previous scenario. 
The Active Memory Dump option is in the same location as the normal dump settings, in the Startup 
And Recovery dialog box of the System Properties, as shown in Figure 2-36. 

---

38 
CHAPTER 2 | Software-defined datacenter 
Figure 2-36:
Active Memory Dump settings 
Note On the topic of memory dumps in general, failover clustering has been integrated with Live 
Dump to capture dumps when timeouts are reached. You then can analyze these dumps for root 
cause. This also integrates with Windows Error Reporting to allow for retention and inclusion of 
other logs. 
Depending on how a resource times out and how failover clustering is configured, Live Dump can 
trigger the machine to issue a stop error (i.e., blue screen) and cause the machine to crash in order to 
create a memory dump. Although the memory dump is useful in determining the cause of a problem, 
it does incur downtime while the machine reboots to create the dump. However, with the integration 
of Live Dump, a memory dump is created in the background while the machine itself continues to run, 
which does not affect production. 
The focus of this feature is to collect enough logs/dumps for Microsoft support to more successfully 
troubleshoot various issues that customers might experience when using clusters. The goal is to 
gather enough information so that support can help solve the problem when customers call, instead 
of asking customers to reproduce the problem and then waiting for a time when it can be done.