Is it normal for the vmkdump folder to contain a dumpfile?

[Meathook]

Hi everyone,

It might be a stupid question, but I was wondering about the following:

I was trying to clean up one of our datastores, and came across a Folder called "vmkdump" containing a single file  named something like "xxxxx-xxxxx-xxxxxx-xxxxx.dumpfile". I found out those a dumps of the ESXi Server used when the machine crashes. Now my question is, are those files created when the machine crashes, or is this file created while the Server is running and data is collected until the Server crashes?

I guess in the end what I'm really asking is, can I get rid of that dumpfile, because we don't Need it anymore, or is at least one active dumpfile needed for the Server to function properly?

Thanks in advance,

Fabian Fritsch

[Meathook]

After searching through the documentation a bit more I stumbled across this entry:

http://pubs.vmware.com/vsphere-65/topic/com.vmware.vsphere.storage.doc/GUID-0DF35CC7-5187-4C60-BC3B-...

Am I getting this right, and this .dumpfile is our actual active Core-Dump-Partition and not a single collection of data? If so, I guess everything is fine and dandy and just as it should be, with a single active partition? Still not sure about this whole topic, someone with more knowledge clearing things up would be much appreciated.

[cyberfed2727]

You can view and change the default location where ESXi does its logging dumps.

On your ESXi host go to advanced settings --- Syslog.global.logDir

That will show you where logs will be dropped to (default setting is /scratch/log  aka local partition that is not persistent)

You can change this to point to a datastore and folder, example such as /datastore1/vmlogs

[Meathook]

Hello cyberfed2727,

thank you for taking your time reading my post and writing your reply. I checked the advanced configuration and the current value of Syslog.global.logDir is "[] /scratch/log", so I guess it's still the normal non persistant partition. I checked on another ESXi machine today and was able to find a similar directory and file on this machine too, so I'm thinking this really is the diagnostic Partition that get's automatically created when installing ESXi, according to this article:

VMware vSphere 6.5 Documentation Library

"During a host failure,  ESXi must be able to save diagnostic information to a preconfigured location for diagnostic and technical support purposes. 

Typically, a partition to collect diagnostic information, also called VMkernel core dump, is created on a local storage device during ESXi Installation."

I'm quite sure this is the case here now, additional input is still welcome, though. Also thanks again to cyberfed2727!

With regards,

Fabian Fritsch

[PCTechStream · Accepted solution]

I just got into the same situation;

Folder: "vmdkdump" & File: ""xxxxx-xxxxx-xxxxxx-xxxxx.dumpfile" stuck in one of 16 LUNs I was deleting.

I tried all my tricks but nothing worked out, so this is what I did if anyone needs to get rid of this file in order to remove datastore from a host.

NOTE:

Make sure you delete the ".dumpfile" & "vmkdump" before removing the datastore.

Enable SSH on all the host to find out which one got the "Active value of true and Configured value of true" It is an "active core dump file"

So how to find out?

1. Run the following command from all the host until one of them display "Active value of true and Configured value of true"

EXAMPLE:

[root@K-YOUR HOST:~] esxcli system coredump file list

Path                                                                                                                                                         Active  Configured              Size

-------------------------------------------------------------------------------------------                                              ------   ----------                 ---------

/vmfs/volumes/xxxxxxxx-xxxxxxxx-xxxx-xxxxxxxxxxxx/vmkdump/xxxxx-xxxxx-xxxxxx-xxxxx.dumpfile    true     true              2542796800

So how to remove it?

2. Run the follwing 2 commands from the host that displays "Active value of true and Configured value of true"

esxcli system coredump file set -u

esxcli system coredump file remove -f /vmfs/volumes/xxxxxxxx-xxxxxxxx-xxxx-xxxxxxxxxxxx/vmkdump/xxxxx-xxxxx-xxxxxx-xxxxx.dumpfile

3. Finally delete the folder: "vmkdump" from the datastore & Unmount/Detache the datastore. DONE!

For more info follow this link: Removing the ESXi coredump to file (2090057)

https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=20900...

Raul.

VMware VDI Administrator

http://ITCloudStream.com/

[Meathook]

Hello Raul,

thank you for your response. You confirmed my guess that these files in the vmkdump folder contain coredumps. Was the coredump you just removed from one of your LUNs the only coredump file you found on the ESXi machine? If so, did you create a new one on a different storage? I have decided to keep the active coredump on our storage (we weren't going to disconnect it anyway, this question came up for me when trying to clean up the storage from dead files), since it seems to be good practise to have an active diagnostic partition assigned.

With regards,

Fabian Fritsch

[PCTechStream]

Fabian!  That was the only file I found, it seems like the "dumpfile" was created & placed by the host on one of the datastore. Also, I didn't create a new one after deleted, remember the host will create and select a datastore for that file, so even if you delete it the esxi will re-create it automatically at the next reboot as long as you have the option checked/unchecked. Even more, you can always select a location.

Advanced Setting on the ESXi Host (Configuration > Software > Advanced Settings)

You don't have to delete "dumpfile" if the host already chose a location. In my case I had to because storage upgrade & new datastore creation. See image....  

I hope that answer your Quest!

Raul.

VMware VDI Administrator

http://ITCloudStream.com/

[roconnor]

Same issue, need to clean up a datastore, but I need some of that powercli goodness to get this done on a bunch of hosts 

#PowerCLI Rocks!

<#

    .SYNOPSIS

    PowerCLI get core dump location

    .EXAMPLE

    Connect-viserver <fqdn vcenter>

    .\Get-HostCoreDumpFile.ps1

    .NOTES

    raoconnor 03/10/17

    Note:

    # Core dump can be network, file, partition and vsan, ie:

    $esxcli.system.coredump.network.get()

#>

foreach($vmhost in Get-VMHost){

    $esxcli = Get-EsxCli -VMHost $vmhost.Name

    $vmhost = $esxcli.system.hostname.get().HostName

    Write-Host "$vmhost" -ForegroundColor DarkYellow

    Write-Host "Active coredump file:" $esxcli.system.coredump.file.get().Active

    Write-Host "Configured coredump file:" $esxcli.system.coredump.file.get().Configured

}

# uncomment the following to unconfigure file core dump

<#

foreach($vmhost in Get-VMHost){

    $esxcli = Get-EsxCli -VMHost $vmhost.Name

    $esxcli.system.coredump.file.set(0)

}

#>