Hi,
Since this BZ is opened on the UI side, we first need to ensure whether it is supported from the backend or not.
UI creates "LocalVolumeDiscovery" CR during the StorageSystem deployment steps and based on the output of "LocalVolumeDiscoveryResult" CR (status.discoveredDevices[n].type) it figures out the available devices of particular supported types (currently "disk" and "part").

1. So, my very first question will be, does "LocalVolumeDiscovery"/ and "LocalVolumeDiscoveryResult" CRs themselves support device types "mpath" ??
2. Second doubt is about the "workaround" mentioned in the description. After applying this manual process, was created StorageCluster got healthy and we got 3 OSDs as expected (here I am trying to make sure that rook supports "mpath" devices as well) ??

Thanks.

(In reply to Sanjal Katiyar from comment #2)
> Hi,
> Since this BZ is opened on the UI side, we first need to ensure whether it
> is supported from the backend or not.
> UI creates "LocalVolumeDiscovery" CR during the StorageSystem deployment
> steps and based on the output of "LocalVolumeDiscoveryResult" CR
> (status.discoveredDevices[n].type) it figures out the available devices of
> particular supported types (currently "disk" and "part").
> 
> 1. So, my very first question will be, does "LocalVolumeDiscovery"/ and
> "LocalVolumeDiscoveryResult" CRs themselves support device types "mpath" ??
> 2. Second doubt is about the "workaround" mentioned in the description.
> After applying this manual process, was created StorageCluster got healthy
> and we got 3 OSDs as expected (here I am trying to make sure that rook
> supports "mpath" devices as well) ??
> 
> Thanks.

Please feel free to add someone from the LSO team, in case we are not sure of point 1. mentioned above !

Hi Sanjal,

Thanks for looking at this request.

(In reply to Sanjal Katiyar from comment #2)
> Hi,
> Since this BZ is opened on the UI side, we first need to ensure whether it
> is supported from the backend or not.
> UI creates "LocalVolumeDiscovery" CR during the StorageSystem deployment
> steps and based on the output of "LocalVolumeDiscoveryResult" CR
> (status.discoveredDevices[n].type) it figures out the available devices of
> particular supported types (currently "disk" and "part").

Thanks for clarifying this part, I thought that, after the LocalVolumeDiscovery CR, 
you could define the device types for LocalVolumeSet to be created and that this 
setting was independent of LocalVolumeDiscovery, but I understand that the deviceType,
while creating the storageSystem CR, comes from the LocalVolumeDiscoveryResults only.

> 
> 1. So, my very first question will be, does "LocalVolumeDiscovery"/ and
> "LocalVolumeDiscoveryResult" CRs themselves support device types "mpath" ??

Not yet, I've filled https://issues.redhat.com/browse/OCPBUGS-10884 for this.

> 2. Second doubt is about the "workaround" mentioned in the description.
> After applying this manual process, was created StorageCluster got healthy
> and we got 3 OSDs as expected (here I am trying to make sure that rook
> supports "mpath" devices as well) ??

It is correct.
I can ask for an ODF must-gather to collect more information about this point if you need

Hi @etamir,
Wdyt about this BZ, is it something we should explore from ODF UI and LSO side in future releases or we don't want to support "mpath" device types from the ODF's side and want customers (if any) to use workaround or any other CLI method ??

Hi Eran,

So fix on LSO's LocalVolumeDiscovery side is merged for LSO 4.14 but there is one limitation currently with LocalVolumeDiscoveryResult CR and since UI uses LocalVolumeDiscoveryResult we might have to update UI + documentation accordingly.

So, issue is that: https://issues.redhat.com/browse/OCPBUGS-10884?focusedId=22855634&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-22855634, LocalVolumeDiscoveryResult CR reports unique devices for types "disk" and "part" but in case of "mpath" it could have an overlap (with no way to differentiate as of now).

As an UI solution LSO team's proposal is to: https://issues.redhat.com/browse/STOR-1430, make sure user can only select "disk" or "part" or "disk+part" or "mpath" from the "Device type" filter dropdown.

Are you okay to move forward and update ODF UI with this, if so do you have any preference whether to pick this up for ODF 4.14 or 4.15 release ? Or, in case you see any issue we can loop in someone from LSO team on this BZ as well (please let me know if that's the case).

PR is merged in master, need acks if we are planning to take it for 4.14 release...

Updated "Doc Text", adding need_info on @olakra just to keep doc team in loop.

Changes are while creation of a new StorageSystem of "Create a new StorageClass using local storage devices" Backing storage type,
on "Create local volume set" wizard step,
for "Advanced" filtering option for "Device type" multi-select dropdown.

UI only restriction (because UI used LocalVolumeDiscovery CR): "Mpath" type selection is exclusive and cannot be used with other device types.

sent PR for master, will get it merged and backported to 4.14 asap !

just adding for reference: UI only selects the disks which are in "Available" state and matches one of these types: "disk" or "part" or "mpath".

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (Important: Red Hat OpenShift Data Foundation 4.14.0 security, enhancement & bug fix update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHSA-2023:6832