On Fri, Jul 22, 2016 at 4:48 PM, Ralf Schenk <rs@databay.de> wrote:

Hello,

I also see from the logs that all your Storage-Domains that work are mounted as nfsVersion='V4' but ovirt-nfs.netsec:/ovirt/hosted-engine is mounted as nfsVersion='null'.


Hi Robert,
unfortunately Ralf is right: I reproduced the issue.

The auto-import procedure for the hosted-engine storage domain ignores the nfsVersion parameter and so we don't have a value for that in the engine DB.
On hosted-engine hosts, the agent mounts the hosted-engine storage domain before the engine and so everything is fine since the agent knows that it's nfsv4.

The issue comes with the hosts of that datacenter not involved in hosted-engine: in this case the engine simply tries to mount without the nfsVersion parameter and so, if the NFS server cannot be access over nfsv3, the mount could fail and the host will be declared as not operation.

I opened a ticket to track it:

If you need a quick fix you can:
- fix the configuration of your storage server to allow it to be accessed also over nfsv3
- edit the configuration of the storage connection in the engine DB on the engine VM to add the missing parameter. Something like:
 # sudo -u postgresl psql
 \c engine;
 select * from storage_server_connections;
 UPDATE storage_server_connections SET nfs_version = '4' WHERE connection = 'ovirt-nfs.netsec:/ovirt/hosted-engine';
 commit;
 select * from storage_server_connections;
 

Bye


Am 22.07.2016 um 16:17 schrieb Simone Tiraboschi:
On Fri, Jul 22, 2016 at 3:47 PM, Robert Story <rstory@tislabs.com> wrote:
Hello,

I'm in the process of upgrading from 3.5.x to 3.6.x. My hosted engine and
hosts in the primary cluster are all upgraded and appear to be running fine.

I have a second cluster of 2 machines which are just regular hosts, without
the hosted-engine. Both have been marked non-operational, with the
following messages logged about every 5 minutes:


Failed to connect Host perses to Storage Pool Default

Host perses cannot access the Storage Domain(s) hosted_storage attached to the Data Center Default. Setting Host state to Non-Operational.

Host perses reports about one of the Active Storage Domains as Problematic.

Failed to connect Host perses to Storage Servers

Failed to connect Host perses to the Storage Domains hosted_storage.


I could see the normal storage/iso/export domains mounted on the host, and
the VMs running on the host are fine.
In 3.5 only the hosts involved in hosted-engine have to access the
hosted-engine storage domain.
With 3.6 we introduced the capabilities to manage the engine VM from
the engine itself so the engine has to import in the hosted-engine
storage domain.
This means that all the hosts in the datacenter that contains the
cluster with the hosted-engine hosts have now to be able to connect
the hosted-engine storage domain.

Can you please check the ACL on the storage server (NFS or iSCSI) that
you use to expose the hosted-engine storage domain?

I shut down the VMs on one host, put it in maintenance mode, installed 3.6
repo and ran yum update. All went well, but when I activated the host, same
deal.

I've attached the engine log snippet for the activation attempt.

Robert

--
Senior Software Engineer @ Parsons

_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users

_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users

--


Ralf Schenk
fon +49 (0) 24 05 / 40 83 70
fax +49 (0) 24 05 / 40 83 759
mail rs@databay.de
 
Databay AG
Jens-Otto-Krag-Straße 11
D-52146 Würselen
www.databay.de

Sitz/Amtsgericht Aachen • HRB:8437 • USt-IdNr.: DE 210844202
Vorstand: Ralf Schenk, Dipl.-Ing. Jens Conze, Aresch Yavari, Dipl.-Kfm. Philipp Hermanns
Aufsichtsratsvorsitzender: Wilhelm Dohmen


_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users