Dynamix Trim with ESXi (RDM) SSD

[joelones]

I'm running ESXi and am doing RDM (Raw Device Mapping) for an SSD that I'm using as a cache drive? Two questions:

 

1. Can the plugin detect an SSD whilst using RDM?
2. Does RDM'ing impact the process of TRIM in any way?

 

I'm assuming it does not work so enabling it would have no effect, correct?

 

root@Tower:~# hdparm -I /dev/sdk | grep -i TRIM
SG_IO: bad/missing sense data, sb[]:  70 00 05 00 00 00 00 0a 00 00 00 00 20 00 00 c0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

I assume 'cause RDM'ing doesn't allow things like temperature sensors etc to talk to unRAID and whatnot but not sure if it impacts TRIM plugin, probably. Guess I just answered my own question. It's been great talking to myself. I'll probably just try to add it to the controller I'm passing in.

[BobPhoenix]

Make sure your controller supports Trim.  My IBM M1015's (LSI 2008 chipset) don't later LSI's like 2308 chipset (I think) do.  What I want to try with ESXi is using a controller (can be just 2 ports) for the datastore drive and pass through the MB controller to unRAID.  Hasn't worked yet but I didn't have time to experiment and see if I HAD a controller that would work like that or not.

[joelones]

20 hours ago, BobPhoenix said:

Make sure your controller supports Trim.  My IBM M1015's (LSI 2008 chipset) don't later LSI's like 2308 chipset (I think) do.  What I want to try with ESXi is using a controller (can be just 2 ports) for the datastore drive and pass through the MB controller to unRAID.  Hasn't worked yet but I didn't have time to experiment and see if I HAD a controller that would work like that or not.

 

Damn, that's precisely what I'm running, two M1015's flashed in IT mode - guess there's no point running Dynamix's trim plugin...

 

I just removed my SSD in RDM mode and added it directly to one of the M1015s. I was hoping by doing this, I'd get trim support....

 

EDIT: Came across another post somewhere else stating that, "LSI cards in "it" mode will support TRIM if software raid is used", is this the case?

 

https://www.broadcom.com/support/knowledgebase/1211161496937/trim-and-sgunmap-support-for-lsi-hbas-and-raid-controllers

 

How does one verify this on the command line?

[BobPhoenix]

Found post by @johnnie.black that tells you what the SSD needs to support to get trim on an SAS2008 based LSI controller:

SAS2008 based controllers only support trim on SSDs with deterministic read after trim / read zero after trim, most SSDs don't support that, SAS2308 and newer support trim on all SSDs.

[joelones]

51 minutes ago, BobPhoenix said:

Found post by @johnnie.black that tells you what the SSD needs to support to get trim on an SAS2008 based LSI controller:

 

So to clarify the SSD needs to implement both: Deterministic Read After Trim (DRAT) and Deterministic Zeroes After Trim (DZAT)?  or one of the two?

 

root@Tower:/mnt/cache/appdata# hdparm -I /dev/sdn | grep -i trim
	   *	Data Set Management TRIM supported (limit 8 blocks)
	   *	Deterministic read ZEROs after TRIM

 

[BobPhoenix]

1 minute ago, joelones said:

 

So to clarify the SSD needs to implement both: Deterministic Read After Trim (DRAT) and Deterministic Zeroes After Trim (DZAT)?  or one of the two?

 

root@Tower:/mnt/cache/appdata# hdparm -I /dev/sdn | grep -i trim
	   *	Data Set Management TRIM supported (limit 8 blocks)
	   *	Deterministic read ZEROs after TRIM

 

Hope @johnnie.black sees this.  I cannot tell you.

[joelones]

Ok thx. SSD is a cache disk and formatted with xfs.

 

Command completed relatively quick on a new drive. Can't tell if it's working or not, seems so...(I think). The command would prompt an error if not working, I assume...

root@Tower:/mnt/cache/appdata# /sbin/fstrim -v /mnt/cache
/mnt/cache: 646.2 GiB (693857992704 bytes) trimmed

 

[JorgeB]

8 hours ago, joelones said:

So to clarify the SSD needs to implement both: Deterministic Read After Trim (DRAT) and Deterministic Zeroes After Trim (DZAT)?  or one of the two?

To be honest I'm not certain, there's little info about it, it's easy to test though, just run fstrim -v and if you get:

 

FITRIM ioctl failed: Remote I/O error -> not working

29.8 GiB (32001204224 bytes) trimmed -> working

 

 

 

[swagler]

@johnnie.black-  I recently got a LSI SAS 9211-4i. Attached Samsung 840 Pro 512GB that does have deterministic read after trim / read zero after trim. I have an Asus A8N-SLI Premium running Windows 7 Ultimate x64. I have a case opened with LSI Support Tech, he concurs that TRIM is supported flashed in IT Mode, but TrimCheck v0.7 is returning it's not working. Any advice would be greatly appreciated.