Question on failover configuration

Printable View

07-20-2011, 07:08 PM
blindzero

Question on failover configuration

Hi Forum,

some general question on the failover config for iSCSI.

Background:
- 2 Open-E DSS
- 4 NICs each
-- 1 NIC LAN (Management, SMB NAS acccess)
-- 1 NIC replication
-- 2 NICs Storage iSCSI VLAN (MPIO)
-- 2 Virtual IPs running on the 2 Storage VLAN NICs

My questions is about
1) the Network interface for unicast.
In the webcast on the setup of failover (with Xen) it is told to set this on the LAN NIC...but does this make sense? Shouldn't it be better to places this on the NIC for the replication? So the management traffic on the failover config is not going over switches but directly between the two nodes?

2) network of ping nodes
What is your experiences about the network for the ping nodes?
We've chosen the two hosts in the Storage iSCSI VLAN as ping nodes as this network is the relevant for some decision on doing some failover or not...

What do you guys think?

Matthias
07-20-2011, 07:17 PM
Gr-R

1- I see your point, but best practices dictate that the replication path be strictly for replication to maximize speeds.

2- Ping nodes should be equipment that has 100% uptime. Like routers, switches, etc. Choosing hosts could be dangerous should hardware fail. Understandably routers and switches can fail too, but that would be less likely.
07-20-2011, 07:31 PM
blindzero

Quote:

Originally Posted by Gr-R

2- Ping nodes should be equipment that has 100% uptime. Like routers, switches, etc. Choosing hosts could be dangerous should hardware fail. Understandably routers and switches can fail too, but that would be less likely.

Thats why we've chosen our Windows DC servers. And we have two, and as I read both ping nodes must fail to perform a failover?
I was against switches as I wanted to put the ping node into the iSCSI VLAN and the switche would have to have an IP in that network then...
Or maybe it was wrong and I should take ping nodes in the regular LAN? Then I could take switches, too. :confused:

Matthias
07-20-2011, 07:41 PM
Gr-R

Either way, uptime is important.
Only those network paths which are connected to storage clients need to be monitored with Ping-Node(s).
07-21-2011, 07:57 AM
Oer2001

No local backup and nfs failover

Hello,

Error 102: Shares used in backup task are on volume which are running in the NAS (NFS) Failover service.

The question is, how I have to do backups of such a share??

Regards,
Erik
07-22-2011, 10:35 AM
Al-S

Quote:

Originally Posted by Oer2001

Hello,

Error 102: Shares used in backup task are on volume which are running in the NAS (NFS) Failover service.

The question is, how I have to do backups of such a share??

Regards,
Erik

As the error says, the shares are in use with an NFS that running a Failover.

On Shares used in the NFS failover, only NFS and data Replication can be used.
So you need to stop failover to take a backup.
And backup cannot work with NFS auto Failover, this issue is in our LAB for tests, and we may have it in the future.

As only data replication is possible on shares that are in NFS failover is running
( replication is like taking a backup for your data, but in the real time, sure it is recommended to have a backup too for very critical data, but you need to stop
the Failover in that case).

_________________
Alaa Souqi