SR usage on multiple pools

[olivierlambert]

A SR cannot be attached to multiple pools at once.

[Tristis Oris]

@olivierlambert but it attached) bug here?

[olivierlambert]

If you plug a block device on another pool it will lead to corruption because one pool cannot know where the other pool is having host writing to which VM disk.

[Tristis Oris]

@olivierlambert need to block plug operations then.

[Tristis Oris]

yep. VM i migrated from old SR now won't start. Looks VDI's was totaly corrupted at moment of connection.

[Tristis Oris]

@olivierlambert nvm restored it by somehow.

Double mount not a problem for NFS storage?

[Danp]

@Tristis-Oris said in SR usage on multiple pools:

Double mount not a problem for NFS storage?

You can create an NFS SR on each pool using the same destination device / path. It will create a unique subdirectory for each pool, so no issues with corruption because each pool has its own unique SR.

[Tristis Oris]

@Danp so i need to migrate VDIs anyway? inside same SR but to another dirrectory.

[Danp]

@Tristis-Oris Yes, but it isn't the same SR. It is multiple SRs, each in their own subdirectory.

[Tristis Oris]

@Danp yes got it. i'm about physical layer. thanks.

[olivierlambert]

Yes, it's a "definition" question/problem: a SR is the exact place where you store your virtual disk images. On NFS, it's a folder named after the UUID of the SR. As Dan said, if you connect to the same NFS folder, it will create a new folder for each new SR inside. So even if it's the same physical device/NFS share, there's many SRs on it.

For a block based SR, it's different: if you pass a LUN (in ISCSI for example), it will attach it and warn you: either you want to re-attach the previously existing SR (UUID etc) either you erase it completely to start from scratch. There's no "folders" on a block space like a LUN, so it's different.