VMware Cloud Community
srwsol
Hot Shot
Hot Shot
Jump to solution

ESXi 6.5 USB Devices attached to VMs Issues?

Has anyone else experienced any issues with ESXi 6.5 and host USB 3 devices attached to VMs?

I'm seeing messages like this from an Arch Linux VM that has a number of USB 3 hard disks attached to it:

Nov 25 02:07:40 archnfs kernel: usb 1-4.1: reset high-speed USB device number 8 using xhci_hcd

There are quite a number of other error messages associated with it to, and I'm betting that data is being lost.  I'm running a big test copy right now, and when it finishes I'm going to verify the md5sum versus the source to find out for sure. 

Tags (1)
49 Replies
Lagus1
Contributor
Contributor
Jump to solution

Could this work?
I made the change on my host but have still been unable to test the hard drives that I had issues with.

Important information about the new ESXi 6.5 USB driver vmkusb, and the legacy USB drivers (2147650)...

0 Kudos
jUDASmILExAlmig
Enthusiast
Enthusiast
Jump to solution

Hi srwsol

Are you still facing the issue by now? I'd tried with this type of Seagate 8TB hard drive with a Plugable USB3.0 drive dock on my VS6.5 server by connecting the device to an Arch Linux. But I can add the device to the VM and it can be recognized inside the guest. Also I tried with some basic I/O operations by copied some files to the disk and no problem hit during my trial. So would you please let me know more detail about your setup like which USB controller you connect your drive to and which steps made the problem occurred on your side? Thanks.

0 Kudos
MarcHuppert
Enthusiast
Enthusiast
Jump to solution

You need to disable the new vmkusb module in ESXi 6.5 and load the older legacy USB drivers with this command:

esxcli system module set -m=vmkusb -e=FALSE



You must reboot your ESXi host to apply it. It works for me

VCDX #181, VSP, VTSP, VCA, VCP-DCV(2+3+4+5+6+6.5+6.7+2019), VCP-DT, VCP-NV, VCAP(DCA4+5+DCD4+5), VCIX-NV, VCIX-DCV, VCI, vExpert, vEpxert NSX, vExpert VSAN and VCDX
srwsol
Hot Shot
Hot Shot
Jump to solution

Sorry for taking so long to get back to this.  I backed out the update and have been running on the older version, so I haven't had any issues.  I'm going to try to update one of my boxes in the next week or so using the latest version, which came out after I ran the previous tests, and if it doesn't work, then I'll try your suggestion.  By the way, what symptoms were you experiencing MarcHuppert, and which version were you using?

0 Kudos
srwsol
Hot Shot
Hot Shot
Jump to solution

Sorry for taking so long to get back.  Per my other post, I rolled back and didn't do anything else regarding the problem.  I'm going to try again next week with an 8tb drive, but using a different cradle ( https://www.amazon.com/gp/product/B01GYOLGSA/ref=oh_aui_search_detailpage?ie=UTF8&psc=1 ), as that's what I have available here at my place in Florida.  This cradle has been working fine under ESXi 6.0 with several different drives, including the 8tb one, so it should work.  I guess I'm going to have to mount the hard drives read only to start with and see if I can recreate the problem while just reading, as I'm afraid the problem is causing data loss if it happens during writes.  If I can recreate the problem, what debugging information do you suggest I gather besides the vmkernel log and the linux vm logs? 

I do want to get upgraded to ESXi 6.5 as it has the lastest VM hardware version which has some fixes for linux graphics.

0 Kudos
srwsol
Hot Shot
Hot Shot
Jump to solution

I upgraded a different ESXi box and the same thing happened.  When I was copying files on an Arch Linux VM with kernel 4.9.11.1 I was getting this message over and over again:

Feb 25 23:43:27 archnfs kernel: usb 2-2: reset SuperSpeed USB device number 3 using xhci_hcd

I did a hash check on the copied file and it looked like it copied correctly, but there were hundreds of these in the kernel log for a copy of a 4gb file.  I also noticed that the transfer speed was less than half normal.  I then disabled the driver per your suggestion,, rebooted ESXi, and that fixed everything.  Copying was back to full speed and no error messages in the log. 

They really need to better test these drivers before they put them out.  Hopefully they aren't going to remove the old drivers anytime soon.

Thanks for your help.

0 Kudos
mouse_77
Contributor
Contributor
Jump to solution

Hi same here with a OMNIKEY 3021 - esxi 6.5

In the lsusb I see it as omnikey 3121 when try to connect to a ubuntu 17.04 or a debian vm machine, booting it disappear from the connected hardware. On the datastore monitor it seems disconnected

any help?

Really thanks in advance

PS already disabled the new vmkusb module

0 Kudos
mouse_77
Contributor
Contributor
Jump to solution

Nobody?

0 Kudos
TomH201110141
Enthusiast
Enthusiast
Jump to solution

You need to add two line to the VMX-File of your VM:

usb.generic.allowCCID = "TRUE"

usb.analyzer.enable = "TRUE"

Power off VM first and then add this line. Maybe you need to reboot the host as well.

In your case you DON'T need to modify the modules of the host with: "esxcli system module set -m=vmkusb -e=FALSE" <-- there is NO need because you already see the usb device with "lsusb"!

0 Kudos
mouse_77
Contributor
Contributor
Jump to solution

Modified to the original settings esxcli system module set -m=vmkusb -e=TRUE and restarted the host

Modified the vmx file with winscp, assigned the omnikey usb in web ui to the vm and started it...nothing after the boot the omnikey disappear. The strange thing is that the vmx file returns unmodified :smileyshocked:

0 Kudos