Oracle VM and Cdot for iSCSI

[kjcreason]

I perused some of the threads but did not find enough information to feel confident that I had the same issue as the others that warranted the showmounts "patch". The showmounts issued sounded like it might be NFS only? Is that true or false?

Here's the background:

Oracle VM 3.2.2 hosts that have been running the old plugin for OVM 3.1.1 from last fall/summer connected to 7-mode filers.

We used iSCSI for all the repositories instead of NFS (having heard the performance was a little better. Is it? Does anyone have any input on that?)

We are migrated to a new filer system running 8.2P3 in Cdot mode.

I was able to install the new 2.0.2 plugin from TGZ and confirm that it did not break the existing plugin, it had no errors from shell or python script during install.

I have now installed it on all the hosts cleanly (though one node now is stuck in an auto discover mode… but I'll deal with it later).

However, I am unable to discover the new filer as a SAN/iSCSI device.

The error when trying to add the new filer is:

(07/16/2014 04:56:58:555 PM)

OVMAPI_B000E Storage plugin command [storage_plugin_validate] failed for storage server [0004fb00000900006b71822927263100] failed with [com.oracle.ovm.mgr.api.exception.FailedOperationException: OVMAPI_4010E Attempt to send command: dispatch to server: iss-vohm02 failed. OVMAPI_4004E Server Failed Command: dispatch https://?uname?:?pwd?@xxx.xxx.xxx.127:8899/api/3 storage_plugin_validate oracle.netapp.SCSIPlugin.StoragePlugin, Status: OSCPlugin.OperationFailedEx:'NetApp Data ONTAP Failure: failed' [Wed Jul 16 16:56:58 CDT 2014] [Wed Jul 16 16:56:58 CDT 2014]] OVMAPI_4010E Attempt to send command: dispatch to server: iss-vohm02 failed. OVMAPI_4004E Server Failed Command: dispatch https://?uname?:?pwd?@xxx.xxx.xxx.127:8899/api/3 storage_plugin_validate oracle.netapp.SCSIPlugin.StoragePlugin, Status: org.apache.xmlrpc.XmlRpcException: OSCPlugin.OperationFailedEx:'NetApp Data ONTAP Failure: failed' [Wed Jul 16 16:56:58 CDT 2014] [Wed Jul 16 16:56:58 CDT 2014] [Wed Jul 16 16:56:58 CDT 2014]

I used the generic iSCSI storage plugin and was able to see the LUNS presented to the hosts. I deleted it afterward.

I have tried a couple different users on the NetApp including the admin one (not preferred method). I am was configuring the "admin" IP address to be the routable IP address we use to administrate the device.

Thoughts?

[nkarthik]

Hello,

The showmount patch only for NFS specifically if you use oracle generic plugin.

Yes, iSCSI is better than NFS as well as depending on the tuning on all the layers (Hardware, application, Operating system, Connectivity between server and storage, storage).

Regarding Discovery, can you check the /var/log/ovs-agent.log and please send to nkarthik@netapp.com

For vserver/storage virtual machine registration, The LIF(IP address from storage) should have management and data access, because the NetApp storage connect plugin communicate to storage system through ZAPI calls, which required management access and provision the LUNs(physical disk) through data access. we normally recommend the LIF in data access network and include management access.

we will help you to solve your issue.

regards,

Karthikeyan.N

(nkarthik@netapp.com)