HSX node replacement

  • 10 May 2024
  • 1 reply

Badge +1


I have a situation in one of my customers. Case 240503-407 .

Here is the situation : 

the customer start with a 3 nodes HSX cluster with 24x16TB disk

He adds a fourth node, and because of a shortage of 16TB disk availability he goes with 20TB disk (also with 768GB of ram)

Now if we look at the disks on an 16TB node we see :

/dev/sdn                                   17T   13T  3.7T  78% /hedvig/d15

/dev/sdc                                   17T   14T  3.4T  80% /hedvig/d4

/dev/sde                                   17T   13T  3.7T  78% /hedvig/d6

/dev/sdw                                   17T   13T  3.7T  78% /hedvig/d24

/dev/sdd                                   17T   13T  3.8T  78% /hedvig/d5

and if we look at the disks on a 20TB node we see :

/dev/sdt                                   19T   14T  4.3T  77% /hedvig/d22
/dev/sdx                                   19T   14T  4.5T  76% /hedvig/d26
/dev/sde                                   19T   14T  4.4T  77% /hedvig/d7
/dev/sdf                                   19T   14T  4.3T  77% /hedvig/d8
/dev/sds                                   19T   15T  4.0T  79% /hedvig/d21

The support said :

- The disk here already seems to be utilizing the 20 TB that is assigned.
- When checked internally, for hyperscale X, the check on the code side would be " equal to or higher ". This means that 20TB could be utilized fully here, which is seen in the disks as well.
- The CAP of 16TB was for hyperscale 1.5, and since we are on hyperscale X here, you can continue to add nodes with 20TB here.

Does it mean they were formated with all the extra space but the Storage Pool won’t use it ?

Also what is happening if we add more nodes in this cluster with 20TB disks ?

The space will never be used ? 

If we end up with only 20TB disks in all the nodes, the cluster will only use 16TB of disk space ?

Is there a procedure somewhere on how to replace a node inside a cluster ?

Or we just pull the power of the node we want to replace and reimage a new node in this cluster ?

Answer from support ticket 240503-407 :

As much as I want it to be shared with you, there is no documentation available for the replacement of a node.
Only documentation would be to replace a disk here.
You could raise the concern with the accounts team which could help you in a similar scenario.
From the support perspective, I assure that there are no such documentation/ step that is available.


Those it means, when you start a cluster with a specifique disk size, the only way to change it is to start a brand new cluster ?


1 reply

Userlevel 4
Badge +12

Hi @Marco Lachance. 

I’ll check this for you and get back to you, AFAIK disks formatting should all have the same total size within the cluster, although let me double check this in case something has changed more recently.