vSekar Blog

Getting Hands Dirty With Technology

Posts and Views are my Own
You are here: Home

Storage Plugins – PSA, NMP

By

This article is Part of Becoming a VMware Storage Expert series 

Storage Plugins – When you use ESXi with a block device presented using iSCSi or FC, you might have seen or heard terms like PSP, SATP, PSA, NMP and ALUA

While we have already discussed a few basics of ALUA, This post will be discussing what are the Plugins involved how they work. This post is a part of Becoming a VMware Storage Expert series

Pluggable Storage Architecture (PSA)

Starting ESXi 3.5 VMware uses Pluggable Storage Architecture (PSA). PSA enables Storage Vendors to build their own Multipathing Plugin(MPP) using API’s under PSA. This will allow 3rd party vendors to develop their Load balancing, failover mechanisms and plug it into the Architecture.

Availability is critical and redundant paths to storage are essential. One of the key functions of the storage component in vSphere is to provide multipathing (that is, which path a given I/O should use if there are multiple paths) and failover (that is, I/O switching to using another path when a path goes down).

VMware, by default, provides a generic multipathing plug-in (MPP) called the Native Multipathing Plug-in (NMP).

Native Multipathing Plug-in (NMP)

Native Multipathing Plug-in (NMP)NMP works with SATP and PSP to handle Failover and Multipathing and it takes care of

  • Registering logical devices with the PSA framework
  • Receiving input/output (I/O) requests for logical devices it registered with the PSA framework
  • Completes the I/Os and posts completion of the SCSI command block with the PSA framework, which includes the following operations:
  • Selecting the physical path to which it sends the I/O requests
  • Handling failure conditions encountered by the I/O requests
  • Handles task management operations, such as aborts/reset

PSA communicates with NMP for the following operations:

  • Opening/closing logical devices
  • Starting I/O to logical devices
  • Aborting an I/O to logical devices
  • Getting the name of the physical paths to logical devices
  • Getting the SCSI inquiry information for logical devices”

The below image explains how they fit together

PSA NMP Storage Plugins

SATP, PSP and MMP are explained in this article.

Becoming a VMware Storage Expert

By

VMware Storage Expert VMware vSphere APD-PDL VVOL VSAN

How to become a VMware Storage Expert. Let me make it clear this is not a new certification VMware has introduced. This is a series I am starting to help people understand the storage part in the ESXi and implement it better.

I have engaged in several support calls where customers complain that the storage isn’t performing well, very high latency, production environment is down after the LUN’s failed over between the storage processors.

Becoming a VMware Storage Expert means you have to understand and master how ESXi works with Storage. I am starting a series in this blog to explain the components involved when a Datastore is presented (Block or NFS) and mounted to an ESXi, how VVOL works, also will touch-base on vSan.

We will go over all the topics about Storage & availability and explain in detail.

What happens to ESXi when it encounters events like APD-PDL

Few other outage scenarios causing the storage to be unavailable

Here is the Series.

Unknown command: `ntp.server.add’ while setting up 6.7 PSC

By

I followed the user guide to setup NTP on the PSC by running the command ntp .server.add –servers 0.europe.pool.ntp.org and I got this error Unknown command: `ntp.server.add’

Many would have encounter issues with NTP while setting up multiple PSC on the environment.  The common issue I encountered was the time skew. Here is how I resolved it

I got an error: The time on this VMware vCenter Server Appliance is out of synchronization with the external Platform Services Controller with which you are trying to register by xxxx seconds

 

I dug into SSH session and started looking at the date started to compare the date between both nodes. Then I realized this is due to the NTP. One of my hosts was synchronizing with NTP and the other one with ESXi

Here is what I did :

Note: These cmdlets will not work on the shell, try it on bash.

Decide you want to use NTP or sync time with the host (using VMware tools)

 

NTP

Command> timesync.get
Mode: host

Changing the mode to NTP

Command> timesync.set –mode NTP
Command> timesync.get
Mode: NTP

The user guide will ask you to run ntp.server.add and ntp.server.set, however, this commands do not work on bash or shell on 6.7 appliance

Command> ntp.server.add –servers 0.europe.pool.ntp.org
Unknown command: `ntp.server.add’

I set ntp using this command
Command> ntp.set –servers 0.europe.pool.ntp.org

Verified the NTP

Command> ntp.get
Servers: 0.europe.pool.ntp.org

Now when I run timesync.get I should see mode as NTP

Command> timesync.get
Mode: NTP

Host (Use VMware Tools Time Synchronization)

If you want to sync the node time with host

Command> timesync.set –mode host

 

While setting up we do not have options to use the UI to make these changes, so we are stuck with CLI.

 

Alternatively, we can set up NTP from the shell

check the status of the NTP Servers configured by running

ntpq -p

To add ntp server edit /etc/ntp.conf

vi /etc/ntp.conf

add the ntp server FQDN or IP address

use ntpq -p to verify