Upgraded from 4.14 to 5.0.1 - Datastore BASE_PATH depreciated

Hi I am hoping someone can point me in the right direction. I recently upgraded oNeb from 4.14 to 5.0.1 and upon doing the database update it said:

“WARNING: The datastore attribute BASE_PATH has been depreciated. It has been removed from the datastore template”. It then gives me instructions to create a symbolic link in the nodes but when I try it… it says failed to create symbolic link: file exists.

The base_path I was using was /var/lib/one/datastores/nas and it wants to link that to /var/lib/one/datastores but it already seems to exist in it.

One more thing - I haven’t yet run onehost sync if that matters… I’m afraid to keep going at this point.

What can I do? Thank you!
~Michael

Administration opennebula@discoursemail.com writes:

Hello,

Hi I am hoping someone can point me in the right direction. I recently upgraded oNeb from 4.14 to 5.0.1 and upon doing the database update it said:

“WARNING: The datastore attribute BASE_PATH has been depreciated. It has been removed from the datastore template”. It then gives me instructions to create a symbolic link in the nodes but when I try it… it says failed to create symbolic link: file exists.

The base_path I was using was /var/lib/one/datastores/nas and it wants to link that to /var/lib/one/datastores but it already seems to exist in it.

I think you may have a problem to link /var/lib/one/datastores/nas to
/var/lib/one/datastores/ since one is under the other.

It may help you, so here is our setup:

We have all our mount points under /var/lib/one/:

  • /var/lib/one/datastores/ a SAN LUN exposed to all hypervisors

  • /var/lib/one/cluster2-datastores/ a SAN LUN dedicated to a VDC with
    its own hypervisors

Then we have two links:

  • /var/lib/one/datastores/103 -> /var/lib/one/cluster2-datastores/103

  • /var/lib/one/datastores/104 -> /var/lib/one/cluster2-datastores/104

So:

  • all our hypervisors have access to the /var/lib/one/datastores LUN,
    it’s a “common datastores”

  • a VDC has its image and system datastore on a specific SAN LUN,
    exposed only to its own hypervisors

Each datastore mount point is exported by NFS[1] for the frontend:

  • /var/lib/one/datastores is exported by one of all the hypervisors

  • /var/lib/one/cluster2-datastores/ is exported by one of the cluster2
    hypervisors

The frontend mount the NFS shares at the right places for the symlink to work.

Regards.

Footnotes:
[1] I use corosync/pacemaker to have HA NFS

Daniel Dehennin
Récupérer ma clef GPG: gpg --recv-keys 0xCC1E9E5B7A6FE2DF
Fingerprint: 3E69 014E 5C23 50E8 9ED6 2AAD CC1E 9E5B 7A6F E2DF