Flash read cache config question

[fsosson]

Hello,

I would configure flash read cache on one of my ESXi 6u2 hosts.  First, I made sure my local disk is SSD enabled :

Second, if I want to add capacity the list is empty :

Can you please enlighten me on this?

Thanks,

Regards,

Fred

[Finikiez]

Hi!

Check if there is any partition on the disk using 'partedUtil getptbl' command.

If there is one delete it.

[fsosson]

Hi Finikiez,

Thanks for your answer.  When I run: partedUtil getptbl "/vmfs/devices/disks/naa.600508b1001ca7c4c3ab178746a17f00" command hangs and does not give the shell back

I have tried the other option, they crash as well.  According to HP iLO the disks is going well.

Any idea ?

Fred

[fsosson]

After a while, I got this error message:

I am still investigating ...

Regards,

[Finikiez]

Input\output erros mean that there are some issues accessing the disk.

I would recommend to check vmkernel.log

Also be sure that you have the latest FW installed on your local controller.

[fsosson]

I updated local disk controller (HP Smart Array P410i Controller) firmware, this was just a minor upgrade (6.64 to 6.65).

Unfortunately it did not help .  Here is vmkernel.log output :

What is strange is I can access the disk with a Linux live CD, I used parted command to delete partitions without any issue.

[Finikiez]

H:0x1 status means - No connect.

This status is returned if the connection is lost to the LUN. This can occur if the LUN is no longer visible to the host from the array side or if the physical connection to the array has been removed.

VMware KB: Understanding SCSI host-side NMP errors/conditions in ESX 4.x and ESXi 5.x

is it properly seated?

Does SSD have the latest FW?

Anyway this is not an ESXi issue, this is something about the device or FW.

I would also advise to check if you have actual hpsa driver on this ESXi host.

[Finikiez]

But maybe I'm wrong.

What was the command you used to mar the device as SSD?

Did you just copied from the KB VMware KB: Enabling the SSD option on SSD based disks/LUNs that are not detected as SSD by default ?

If so, you attached the wrong SATP rule for your device, as it shouldn't be like VMW_SATP_CX (which is notcied in vmkernel logs)

[fsosson]

For information, firmware and driver are up to date.

I ran: esxcli storage nmp satp rule add --satp=VMW_SATP_CX --device naa.600508b1001ca7c4c3ab178746a17f00 --option "enable_ssd"

When I ran: esxcli storage nmp satp rule add --satp=VMW_SATP_LOCAL --device naa.600508b1001ca7c4c3ab178746a17f00 --option "enable_local enable_ssd" it throws this error message:

     Error adding SATP user rule: Sysinfo error on operation returned status : Bad parameter. Please see the VMkernel log for detailed error information

Here is vmkernel.log output :

What should I check at 'rule' level?

[Finikiez]

You need to use default SATP rule for the device.

This is the first step in KB to check the current SATP rule and use it further.

Currently you need to unclaim the device using 'esxcli storage core claiming unclaim

Then delete created SATP rule esxcli storage nmp satp rule remove

Then reboot your host and repeat steps from KB.

1. Figure our SATP rule 'esxcli storage nmp device list -d naa.id'

2. Using this SATP add satp rule esxcli storage nmp satp rule add --satp=SATP_TYPE --device naa.6006016015301d00167ce6e2ddb3de11 --option "enable_ssd"

where SATP_TYPE is what you got from step 1.

And do all other steps from KB VMware KB: Enabling the SSD option on SSD based disks/LUNs that are not detected as SSD by default


[fsosson]

As soon as I want to remove the existing rule like this, this error message pops in:

This makes me think that there is something wrong somewhere else.

[Finikiez]

Your device is claimed using vmw_satp_cx rule,in your command you used vmw_satp_local.

[fsosson]

sure, I cleaned up everyhting and ran again steps from https://kb.vmware.com/kb/2013188 but 'Add Virtual Flash Resource Capacity' from web GUI is still unfortunately empty.

partedUtil command still displays same error message:

[Finikiez]

just to be sure,

could you paste output from the command

esxcli storage nmp device list -d naa.600508b1001ca7c4c3ab178746a17f00

[fsosson]

certainly, please:

[Finikiez]

Thanks!

You see that SATP rule is VMW_SATP_CX. However this is SATP rule for EMC Clariion. This can causes the issue.

If you can, I would recommend to reinstall ESXi host just to revert all changes. Then start again the process marking the device as SSD.

First after reinstall we need to check the correct SATP rule by running command

esxcli storage nmp device list -d naa.600508b1001ca7c4c3ab178746a17f00

This will let us know for which SATP rulle we should add option 'enable_ssd'

[fsosson]

don't you think I can change that without reinstalling ESXi ?

[Finikiez]

Of course you can.

But when you revert settings and claim your ssd device with correct SATP rule, repeat steps to mark is as ssd and use _correct_ SATP .

If you get an error, post it here so we could check it.

[fsosson]

same message like before

it's like there is a hardware issue but I see the problem on another ESXi host (same hardware), it's HP ProLiant BL460c G7

[Finikiez]

Could you check current SATP rule for the device

esxcli storage nmp device list -d naa.600508b1001ca7c4c3ab178746a17f00

If it's still VMW_SATP_CX your command won't work.