We have an exciting update for the VMware Community today with the release of the Community NVMe Driver for ESXi Fling v1.1. After the release of ESXi 6.7, a number of consumer NVMe devices were no longer bering detected which had previously worked with ESXi 6.5.
One workaround that many folks had to implement, including myself, was to replace the 6.7 NVMe driver. with the previous 6.5 version of the NVMe driver. Although this method still works with the latest ESXi 7.0 Update 2, it is less than ideal as you are using an extremely old version of the NVMe driver and you are giving up new NVMe device enablement, bug fixes and enhancements with the latest version of the driver.
Given the amount of comments and feedback that I had received in the previous blog post, I wanted to explore a way in which we can enable some of these consumer NVMe devices for the community. With the help from Yibo, we have updated the Community NVMe Driver for ESXi Fling to include support for the following storage devices:
| VendorName | VendorId | DeviceId |
|---|---|---|
| ADATA | 0x1cc1 | 0x8201 |
| Micro/Crucial | 0xc0a9 | 0x2263 |
| Silicon Motion/Transcend | 0x126f | 0x2262 |
With this update, we have also added support for ESXi 7.0 or greater when using the new driver which should hopefully be a welcome update for VMware Homelabs!
Here is a screenshot of my Intel NUC 9 running ESXi 7.0 Update 2 which now recognizes my Crucial 1TB NVMe (CT1000P1SSD8)
To help others in the community, if folks want to share the specific brand and model of their NVMe device that can now be enabled by the new driver, feel free to leave a comment. Other consumer NVMe devices may be considered in the future, but will soley be based on feedback from the community. If your particular NVMe device is not being picked up, please share the vendor/device ID by using lspci -v command on ESXi Shell.
Oooh, this is awesome. Excellent work team.
Now if Realtek could come to the party and make a native driver for their NICs.
Will this fling support Patriot 512GB NMVe module? I have 4 of them in 4 different ESXi boxes and had to downgrade to the 6.5 driver to make it work.
Take a look at the Vendor/DeviceID as mentioned in blog post to see if your device is included 🙂
unbelievable - I really appreciate the effort you put into making this work again! Great to be able to update properly again 🙂
Thanks for the great update!
It worked fine with the ESXi 7.0u2 and Transcend NVMe SSD.
Driver :
nvme-community-driver_1.0.1.0-2vmw.700.1.0.15843807-component-18290856.zip
Tested SSD :
Transcend PCIe SSD 220S TS2TMTE220S
* controller: Silicon Motion, VID:0x126f, PID:2262
----
[[email protected]:~] lspci -v
0000:3a:00.0 Mass storage controller Non-Volatile memory controller: Silicon Motion, Inc. Device 2262 [vmhba1]
Class 0108: 126f:2262
[[email protected]:~] vdq -q
[
{
"Name" : "t10.NVMe____TS2TMTE220S_____________________________G095430113__________00000001",
"VSANUUID" : "",
"State" : "Ineligible for use by VSAN",
"Reason" : "Has partitions",
"IsSSD" : "1",
"IsCapacityFlash": "0",
"IsPDL" : "0",
"Size(MB)" : "1953514",
"FormatType" : "512e",
"IsVsanDirectDisk" : "0",
},
]
PLEXTOR PX-2TM10PY
0000:02:00.0 Mass storage controller Non-Volatile memory controller: Vendor 1e95 Device 1005 [vmhba1]
Class 0108: 1e95:1005
Unfortunately I did not recognize.
Silicon Motion / TEAMGROUP (MP33 M.2 PCIe SSD)
0000:01:00.0 Mass storage controller Non-Volatile memory controller: Silicon Motion, Inc. Device 2263 [vmhba1]
Class 0108: 126f:2263
from [email protected]
Hello William,
Unfortunately after trying to update to 7.0u2a the storage was not recognized for my servers. Details below for the storage controller
0000:01:00.0 Mass storage controller Mass storage controller: Apple Inc. ANS2 NVMe Controller [vmhba1]
Class 0180: 106b:2005
Hi William, thanks for this. Actually this is an Intel 660p 1Tb, not sure why it's showing up as 2Tb?
0000:05:00.0 Mass storage controller Non-Volatile memory controller: Intel Corporation SSDPEKNW020T8 [660p, 2TB] [vmhba4]
Class 0108: 8086:f1a8
William, did you ever run into an issue with your CT1000P1SSD8 where it seems to just disconnect from ESXI whenever it hits some mysterious threshold of throughput? Happens to me on 7.0 (up to update, tested on most releases) using vSAN or just dumping files on it directly attached to test. Reboot brings it back, but it will drop again whenever I hit a high iops on it again. I have four of them on two r720s, all do the same thing.
While I have the same device and IDs as you - mine is recognized without this community driver, and.. I get the same behavior with it.
Ran into an issue with a Seagate Barracuda 510 1Tb NVMe on ESXi 6.5 U3, where if it hit its max temp threshold (75 deg) ESXi would simply detach the storage. I expected it to just throttle itself like my Samsung 970 EVO. Since using a proper heatsink it never goes over 55 deg
That was exactly it, found a similar comment right after I wrote this. I came back to comment on that for anyone else who shares the problem. After putting heatsinks on them the issue went away.
Samsung PM961, 144d:a804