MSA206x for VMware best practices

[CadenLange]

Hi

I've read the HPE best practices guide for 6th Gen MSAs (we've a MSA2062 with ADS) and followed all the best practices included. I noticed there isn't a VMware specific best practices guide for 6th Gen MSA (not that I can see) but I did find one for 5th Gen (MSA 2052 etc). This included a couple of specific VMware changes relating to setting 'missing lun response' and IOPs=1 for RR multipathing etc

Are these still relevent for 6th Gen arrays when running VMware 6.7? We're soon to be upgrading to 7.0 U3. Also we only have 2 hosts that are directly attached using SAS and a single DG on controller A (so controller B is passive). Although we set RR PSP for all volumes there is only 1 active path therefore I'm assuming the IOPs=1 setting is pretty irrelevant.

thanks in advance.

[Mahesh202]

Hi CadenLange

While VMware-specific best practices guides can be helpful, many of the best practices outlined in the HPE best practices guide for 6th Gen MSAs are also applicable to VMware environments.

Regarding the specific changes you mentioned from the 5th Gen MSA VMware guide, the "missing lun response" setting is still applicable for 6th Gen MSAs. This setting specifies how the ESXi host should respond if it encounters a missing LUN. The recommended setting is "failover," which allows the host to switch to the alternate path if the primary path is missing.

However, the "IOPs=1" setting for round-robin multipathing is not applicable in your configuration since you only have one active path. This setting is designed to limit the number of I/O operations sent to a single path to prevent overloading it. With only one active path, there is no need to limit the IOPs.

When upgrading to VMware 7.0 U3, it is always a good idea to review the latest VMware best practices guides and HPE documentation to ensure that you are following the most up-to-date recommendations for your environment.

Hope this helps.!!

Regards
Mahesh.

[CadenLange]

Many thanks Mahesh
My only follow up question on the details you’ve provided is regarding the recommended setting for the missing lun response. The HPE best practices guide for VMware environments for 5th generation MSAs states this setting should be configured with a value of ‘illegal’ whereas you’re suggesting a value of ‘failover’. Is that a change to the HPE best practice for VMware environments for 6th generation MSAs only or should this setting be applied to previous MSA generations too?

Kind regards

[Mahesh202]

Hi CadenLange

Thank you for bringing this to my attention. I apologize for any confusion I may have caused. Upon further research, I found that the recommended setting for the "missing LUN response" may vary depending on the specific scenario.

In general, the recommended setting for the "missing LUN response" is "failover" when using round-robin multipathing. This allows the ESXi host to switch to the alternate path if the primary path is missing.

In any case, I would recommend reviewing the latest HPE documentation and VMware best practices guides for your specific MSA model and VMware version to ensure you are following the most up-to-date recommendations for your environment. Additionally, if you have any questions or concerns, it is always recommended to contact HPE Tech support for assistance.

Here's the Best Practices guide link: https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=a00115998en_us

Hope this helps.!!

Regards
Mahesh.

[CadenLange]

@Mahesh202 Thanks Mahesh

In a way that circles me round to my oringinal question - There is no mention of this setting in the MSA1060 best practices guide you linked too..... but you do say it's still applicable. I'm wondering if it's now the default setting and doesn't need to be adjusted - as otherwise it would address this requirement within the published best practice guide.

I'll assume the best practices guide covers all the best practices for the 6th gen arrays and as setting the lun reponse has been subsequently removed from the 6th gen best practices I will assume it is no longer a best pratice (or perhaps not necessary) to make changes to this particular setting.

Kind regards

[Mahesh202 · Accepted solution]

Hi CadenLange

I found some more information on the " Missing LUN Response " Issue.

Some operating systems do not look beyond LUN 0 if they do not find a LUN 0 or cannot handle noncontiguous
LUNs. The Missing LUN Response option handles these situations by enabling the host drivers to continue probing for
LUNs until they reach the LUN to which they have access.
This option controls the SCSI sense data returned for volumes that are not accessible because they don’t exist or have
been hidden through volume mapping (this does not apply to volumes of offline vdisks). Use the default value, Not
Ready, unless the system is used in a VMware environment or a service technician asks you to change it to work
around a host driver problem.

To change the missing LUN response:

1. In the Configuration View panel, right-click the system and select Configuration > Advanced Settings > Cache.
2. Set the Missing LUN Response option to either:
• Not Ready. Sends a reply that there is a LUN where a gap has been created but that it’s “not ready.” Sense
data returned is a Sense Key of 2h and an ASC/ASCQ of 04/03. This option is the default.
• Illegal Request. Sends a reply that there is a LUN but that the request is “illegal.” Sense data returned is a
Sense Key of 5h and an ASC/ASCQ of 25/00. If the system is used in a VMware environment, use this
option.
3. Click Apply.

Hope this helps.!!

Regards
Mahesh

[Sunitha_Mod]

Hello @CadenLange,

We do see that you have marked the topic as solved. We are glad to know your concern has been addressed. 

[eromanov]

Dear Mahesh202 and others!

I need some advice regarding modifying LUN Response settings on an existing setup. Specifically, I'm looking at changing the LUN Response to ILLEGAL REQUEST on an HPE MSA Storage array, as recommended by HPE for use with VMware ESXi hosts. However, I have some concerns because the documentation always seems to mention doing this "before creating volumes."

My Question: Is it safe to change the LUN Response setting on LUNs that already have data on them and are currently in use by ESXi hosts? What are the risks involved, if any?

Here's my current setup:

• I have active datastores on multiple LUNs.
• ESXi version is 7.0.3, build 23794027.
• Datastores are accessed via iSCSI, and HPE recommends setting the missing-lun-response to ILLEGAL.
• HPE documentation mentions shutting down the hosts before making the change, but it’s not very clear on how changing this setting affects datastores that are already hosting live virtual machines.

My biggest concern is whether this setting can be changed without affecting data integrity or causing potential downtime.

Thanks in advance for any advice you can offer!

[Mahesh202]

Hi Eromanov

May I request you to have a look KB article using the below link

here's link "https://support.hpe.com/hpesc/public/docDisplay?docId=kc0126071en_us&page=GUID-40D4E138-70CC-4D5A-9E28-B150431B2404.html&docLocale=en_US"

also Please go through HP MSA 2040 Storage Configuration and Best Practices for VMware vSphere using this link: http://www8.hp.com/h20195/v2/GetDocument.aspx?docname=4AA4-7060ENW

Hope this helps.!! 

Regards
Mahesh