• Skip to primary navigation
  • Skip to main content
  • Skip to primary sidebar

Yellow Bricks

by Duncan Epping

  • Home
  • ESXTOP
  • Stickers/Shirts
  • Privacy Policy
  • About
  • Show Search
Hide Search

stretched

VMs which are not stretched in a stretched cluster, which policy to use?

Duncan Epping · Dec 14, 2020 · 2 Comments

I’ve seen this question popping up regularly. Which policy setting (“site disaster tolerance” and “failures to tolerate”) should I use when I do not want to stretch my VMs? Well, that is actually pretty straight forward, in my opinion, you really only have two options you should ever use:

  • None – Keep data on preferred (stretched cluster)
  • None – Keep data on non-preferred (stretched cluster)

Yes, there is another option. This option is called “None – Stretched Cluster” and then there’s also “None – Standard Cluster”. Why should you not use these? Well, let’s start with “None – Stretched Cluster”. In the case of “None – Stretched Cluster”, vSAN will per object decide where to place it. As you hopefully know, a VM consists of multiple objects. As you can imagine, this is not optimal from a performance point of view, as you could end up having a VMDK being placed in Site A and a VMDK being placed in Site B. Which means it would read and write from both locations from a storage point of view, while the VM would be sitting in a single location from a compute point of view. It is also not very optimal from an availability stance, as it would mean that when the intersite link is unavailable, some objects of the VM would also become inaccessible. Not a great situation. What would it look like? Well, potentially something like the below diagram!

Then there’s “None – Standard Cluster”, what happens in this case? When you use “None – Standard Cluster” with “RAID-1”, what is going to happen is that the VM is configured with FTT=1 and RAID-1, but in a stretched cluster “FTT” does not exist, and FTT automatically will become PFTT. This means that the VM is going to be mirrored across locations, and you will have SFTT=0, which means no resiliency locally. It is the same as “Dual Site Mirroring”+”No Data Redundancy”!

In summary, if you ask me, “none – standard cluster” and “none – stretched cluster” should not be used in a stretched cluster.

Can I still provision VMs when a vSAN Stretched Cluster site has failed? Part II

Duncan Epping · Dec 18, 2019 · Leave a Comment

3 years ago I wrote the following post: Can I still provision VMs when a VSAN Stretched Cluster site has failed? Last week I received a question on this subject, and although officially I am not supposed to work on vSAN in the upcoming three months I figured I could test this in the evening easily within 30 minutes. The question was simple, in my blog I described the failure of the Witness Host, what if a single host fails in one of the two “data” fault domains? What if I want to create a snapshot for instance, will this still work?

So here’s what I tested:

  • vSAN Stretched Cluster
  • 4+4+1 configuration
    • Meaning, 4 hosts in each “data site” and a witness host, for a total of 8 hosts in my vSAN cluster
  • Create a VM with cross-site protection and RAID-5 within the location

So I first failed a host in one of the two data sites. When I fail the host, the following is what happens when I create a VM with RAID-1 across sites and RAID-5 within a site:

  • Without “Force Provisioning” enabled the creation of the VM fails
  • When “Force Provisioning” is enabled the creation of the VM succeeds, the VM is created with a RAID-0 within 1 location

Okay, so this sounds similar to the originally described scenario, in my 2016 blog post, where I failed the witness. vSAN will create a RAID-0 configuration for the VM. When the host returns for duty the RAID-1 across locations and RAID-5 within each location is then automatically created. On top of that, you can snapshot VMs in this scenario, the snapshots will also be created as RAID-0. One thing to mind is that I would recommend removing “force provisioning” from the policy after the failure has been resolved! Below is a screenshot of the component layout of the scenario by the way.

I also retried the witness host down scenario, and in that case, you do not need to use the “force provisioning” option. One more thing to note. The above will only happen when you create a RAID configuration which is impossible to create as a result of the failure. If 1 host fails in a 4+4+1 stretched cluster you would like to create a RAID-1 across sites and a RAID-1 within sites then the VM would be created with the requested RAID configuration, which is demonstrated in the screenshot below.

Can you move a vSAN Stretched Cluster to a different vCenter Server?

Duncan Epping · Sep 17, 2019 · 2 Comments

I noticed a question today on one of our internal social platforms, the question was if you can move a vSAN Stretched Cluster to a different vCenter Server. I can be short, I tested it and the answer is yes! How do you do it? Well, we have a great KB that actually documents the process for a normal vSAN Cluster and the same applies to a stretched cluster. When you add the hosts to your new vCenter Server and into your newly created cluster it will pull in the fault domain details (stretched cluster configuration details) from the hosts itself, so when you go to the UI the Fault Domains will pop up again, as shown in the screenshot below.

What did I do? Well in short, but please use the KB for the exact steps:

  • Powered off all VMs
  • Placed the hosts into maintenance mode (do not forget about the Witness!)
  • Disconnected all hosts from the old vCenter Server, again, do not forget about the witness
  • Removed the hosts from the inventory
  • Connected the Witness to the new vCenter Server
  • Created a new Cluster object on the new vCenter Server
  • Added the stretched cluster hosts to the new cluster on the new vCenter Server
  • Took the Witness out of Maintenance Mode first
  • Took the other hosts out of maintenance

That was it, pretty straight forward. Of course, you will need to make sure you have the storage policies in both locations, and you will also need to do some extra work if you use a VDS. Nevertheless, it works pretty much straight-forward and as you would expect it to work!

All-Flash Stretched vSAN Cluster and VM/Host Rules

Duncan Epping · Aug 20, 2018 ·

I had a question last week about the need for DRS Rules or also known as VM/Host Rules in an All-Flash Stretched vSAN Infrastructure. In a vSAN Stretched Cluster, there is read locality implemented. The read locality functionality, in this case, ensures that: reads will always come from the local fault domain.

This means that in the case of a stretched environment, the reads will not have to traverse the network. As the maximum latency is 5ms, this avoids a potential performance hit of 5ms for reads. The additional benefit is also that in a hybrid configuration we avoid needing to re-warm the cache. For the (read) cache re-warming issue we recommend our customers to implement VM/Host Rules. These rules ensure that VMs always run on the same set of hosts, in a normal healthy situation. (This is explained on storagehub here.)

What about an All-Flash cluster, do you still need to implement these rules? The answer to that is: it depends. You don’t need to implement it for “read cache” issues, as in an all-flash cluster there is no read cache. Could you run without those rules? Yes, you can, but if you have DRS enabled this also means that DRS freely moves VMs around, potentially every 5 minutes. This also means that you will have vMotion traffic consuming the inter-site links, and considering how resource hungry vMotion can be, you need to ask yourself if cross-site load balancing adds anything, what the risk is, what the reward is? Personally, I would prefer to load balance within a site, and only go across the link when doing site maintenance, but you may have a different view or set of requirements. If so, then it is good to know that vSAN and vSphere support this.

Isolation Address in a 2-node direct connect vSAN environment?

Duncan Epping · Nov 22, 2017 ·

As most of you know by now, when vSAN is enabled vSphere HA uses the vSAN network for heartbeating. I recently wrote an article about the isolation address and relationship with heartbeat datastores. In the comment section Johann asked what the settings should be for 2-Node Direct Connect with vSAN. A very valid question as an isolation is still possible, although not as likely as with a stretched cluster considering you do not have a network switch for vSAN in this configuration. Anyway, you would still like the VMs that are impacted by the isolation to be powered off and you would like the other remaining host to power them back on.

So the question remains, which IP Address do you select? Well, there’s no IP address to select in this particular case. As it is “direct connect” there are probably only 2 IP addresses on that segment (one for host 1 and another for host 2). You cannot use the default gateway either, as that is the gateway for the management interface, which is the wrong network. So what do I recommend:

  • Disable the Isolation Response >> set it to “leave powered on” or “disabled” (depends on the version used
  • Disable the use of the default gateway by setting the following HA advanced setting:
    • das.usedefaultisolationaddress = false

That probably makes you wonder what will happen when a host is isolated from the rest of the cluster (other host and the witness). Well when this happens then the VMs are still killed, but not as a result of the isolation response kicking in, but as a result of vSAN kicking in. Here’s the process:

  • Heartbeats are not received
  • Host elects itself master
  • Host pings the isolation address
    • If the host can’t ping the gateway of the management interface then the host declares itself isolated
    • If the host can ping the gateway of the management interface then the host doesn’t declare itself isolated
  • Either way, the isolation response is not triggered as it is set to “leave powered on”
  • vSAN will now automatically kill all VMs which have lost access to its components
    • The isolated host will lose quorum
    • vSAN objects will become isolated
    • The advanced setting “VSAN.AutoTerminateGhostVm=1” allows vSAN to kill the “ghosted” VMs (with all components inaccessible).

In other words, don’t worry about the isolation address in a 2-node configuration, vSAN has this situation covered! Note that “VSAN.AutoTerminateGhostVm=1” only works for 2-node and Stretched vSAN configurations at this time.

UPDATE:

I triggered a failure in my lab (which is 2-node, but not direct connect), and for those who are wondering, this is what you should be seeing in your syslog.log:

syslog.log:2017-11-29T13:45:28Z killInaccessibleVms.py [INFO]: Following VMs are powered on and HA protected in this host.
syslog.log:2017-11-29T13:45:28Z killInaccessibleVms.py [INFO]: * ['vm-01', 'vm-03', 'vm-04']
syslog.log:2017-11-29T13:45:32Z killInaccessibleVms.py [INFO]: List inaccessible VMs at round 1
syslog.log:2017-11-29T13:45:32Z killInaccessibleVms.py [INFO]: * ['vim.VirtualMachine:1', 'vim.VirtualMachine:2', 'vim.VirtualMachine:3']
syslog.log:2017-11-29T13:46:06Z killInaccessibleVms.py [INFO]: List inaccessible VMs at round 2
syslog.log:2017-11-29T13:46:06Z killInaccessibleVms.py [INFO]: * ['vim.VirtualMachine:1', 'vim.VirtualMachine:2', 'vim.VirtualMachine:3']
syslog.log:2017-11-29T13:46:06Z killInaccessibleVms.py [INFO]: Following VMs are found to have all objects inaccessible, and will be terminated.
syslog.log:2017-11-29T13:46:06Z killInaccessibleVms.py [INFO]: * ['vim.VirtualMachine:1', 'vim.VirtualMachine:2', 'vim.VirtualMachine:3']
syslog.log:2017-11-29T13:46:06Z killInaccessibleVms.py [INFO]: Start terminating VMs.
syslog.log:2017-11-29T13:46:06Z killInaccessibleVms.py [INFO]: Successfully terminated inaccessible VM: vm-01
syslog.log:2017-11-29T13:46:06Z killInaccessibleVms.py [INFO]: Successfully terminated inaccessible VM: vm-03
syslog.log:2017-11-29T13:46:06Z killInaccessibleVms.py [INFO]: Successfully terminated inaccessible VM: vm-04
syslog.log:2017-11-29T13:46:06Z killInaccessibleVms.py [INFO]: Finished killing the ghost vms

 

** Disclaimer: This article contains references to the words master and/or slave. I recognize these as exclusionary words. The words are used in this article for consistency because it’s currently the words that appear in the software, in the UI, and in the log files. When the software is updated to remove the words, this article will be updated to be in alignment. **

  • Go to page 1
  • Go to page 2
  • Go to page 3
  • Go to page 4
  • Go to Next Page »

Primary Sidebar

About the author

Duncan Epping is a Chief Technologist in the Office of CTO of the HCI BU at VMware. He is a VCDX (# 007) and the author of multiple books including "vSAN Deep Dive" and the “vSphere Clustering Technical Deep Dive” series.

Upcoming Events

04-Feb-21 | Czech VMUG – Roadshow
25-Feb-21 | Swiss VMUG – Roadshow
04-Mar-21 | Polish VMUG – Roadshow
09-Mar-21 | Austrian VMUG – Roadshow
18-Mar-21 | St Louis Usercon Keynote

Recommended reads

Sponsors

Want to support us? Buy an advert!

Advertisements

Copyright Yellow-Bricks.com © 2021 · Log in