Solved

PC Scaleout migration options

  • 17 November 2019
  • 6 replies
  • 1980 views

Userlevel 1
Badge +12

I just deployed Prism central scale out on my ESXi cluster and see that the two additional PC appliances have an iso mounted to their CDROM drives. Does this need to stay connected? I’d like to spread these 3 appliances across different datastores. 

(And ideally, I’d like to move these across sites but I understand that this is a future feature: https://next.nutanix.com/product-ideas-61/multi-cluster-scale-out-prism-central-28305)

icon

Best answer by Alona 19 November 2019, 01:38

@mandg 

Thank you for the answers and clarification.

I’d probably agree to the suggestion of raising a case since your questions while being general are relevant to your specific environment.

The task is progressing which is a good thing and without understanding the workload on the cluster it is hard to tell whether it is too slow or on track.

The message you are seeing is relevant to the progressing task. Once the task completes so should the message disappear.

View original

This topic has been closed for comments

6 replies

Userlevel 5
Badge +5

@mandg 

You can unmount the ISO, it is not required for normal operations.

Let me know how you go, please.

Userlevel 1
Badge +12

Thanks, Alona.

I attempted to disconnect the iso and recieved this message in vCenter;

The guest operating system has locked the CD-ROM door and is probably using the CD-ROM, which can prevent the guest from recognizing media changes. If possible, eject the CD-ROM from inside the guest before disconnecting. Disconnect anyway and override the lock?

I responded with ‘No’. For whatever it’s worth, I see the following active Task in PC that has stayed at 83% for a while...; 

"Changing metadata fault tolerance from 0 to 1" with a subtask "Synchronizing metadata replicas"

Userlevel 5
Badge +5

@mandg

OK, so the two important questions are:

  • are PC VMs active and working?
  • are you able to migrate PC VMs between the datastores as you intended?

Also, upon further research I was able to find that the ISO is the only boot source for scaled PC to boot from hence you would need to mount the ISO back to PC VMs post migration. My apologies for any inconvenience caused.

"Changing metadata fault tolerance from 0 to 1" with a subtask "Synchronizing metadata replicas"  

This could be part of PC scale out and is normal provided that it completes eventually. Changing fault tolerance from none to FT=1 is a good thing.

Out of curiosity, would you please share the reasoning behind having more then one storage container (datastore)?

Userlevel 1
Badge +12

Thank you for confirming that boot iso is required to stay with the additional PC appliances. I have a few follow up questions but first I’ll answer yours;

  • Yes, all 3 PC’s are active and working. 
  • I am able to migrate PC from one datastore to another however the iso file stays in the original datastore. I presume that I’d need to power down the appliance since the seems to be locked when it’s running, unmount the iso file, migrate the iso to the destination datastore,  remount it to the appliance, and then power it back up. 
  • Lastly, I don’t really have a good reason to move these to a different datastore within the same cluster, however, I would ultimately like to separate these PC appliances across sites and on different AOS clusters. Do you know if this feature is still on the roadmap for PC Scaleout?

Now my questions are; 

  1. Strangely, one of the 2 new PC appliances has multiple folders that host its vmdk disk files. I’m not sure how or why that happened but it’s only on this one PC appliance.
  2. The "Changing metadata fault tolerance from 0 to 1"  Task is now at 88% - so it’s only progressed 5% in the past 9 hours or so.
  3. On the Prism Central Management page, I still see the message;

     “Prism central scale out currently in progress. Check the tasks page for details.”

 Let me know if you’d prefer that I open a support case on these. Thanks!!

Userlevel 5
Badge +5

@mandg 

Thank you for the answers and clarification.

I’d probably agree to the suggestion of raising a case since your questions while being general are relevant to your specific environment.

The task is progressing which is a good thing and without understanding the workload on the cluster it is hard to tell whether it is too slow or on track.

The message you are seeing is relevant to the progressing task. Once the task completes so should the message disappear.

Badge +1

Thanks, Alona.

I attempted to disconnect the iso and recieved this message in vCenter;

The guest operating system has locked the CD-ROM door and is probably using the CD-ROM, which can prevent the guest from recognizing media changes. If possible, eject the CD-ROM from inside the guest before disconnecting. Disconnect anyway and override the lock?

I responded with ‘No’. For whatever it’s worth, I see the following active Task in PC that has stayed at 83% for a while...; 

"Changing metadata fault tolerance from 0 to 1" with a subtask "Synchronizing metadata replicas"

The message - "Changing metadata fault tolerance from 0 to 1" with a subtask "Synchronizing metadata replicas” is expected as it is copying the metadata from single PC to 2 other VMs which are deployed. If you run the command nodetool -h 0 ring , you would see the size of the metadata increasing for other 2 PC/CVMs. Also yes you can have all 3 PC vms in diff containers from the same cluster, but there is no benefit as it is coming from the same storage pool.