2011-02-16 07:37 AM
Does anyone configured a Fabric MetroCluster without usin syncmirror?
Is that supported??
our customer bought a Metrocluster gateway, but they have just one disk cabin at this moment.
1) I would like if it is a supported configuration / will it work?
2) if not, can i implement syncmirror taking plexes from the same box ( altough it is senseless; syncmirror intends to have physically separated disks )???
Thanks in advance and best regards, Alberto
2011-02-16 10:36 PM
There is little point in having metrocluster without syncmirror; it makes it impossible to take over in case of site outage. For the same reason “syncmirror in a box” it totally pointless.
I do not understand what “disk cabin” is. Could you elaborate?
2011-02-17 04:22 AM
Thanks for the post.
I already know this configuration is absolotely pointless, i just asking about if it was possible.
I found this Note in the MC implementation guide.
Make sure all storage is set up under SyncMirror. While nonmirrored storage is technically permissible, NetApp does not recommend it, since the data on that storage is not available after a site failover.
This is, my configuration is not recommended ( as it is supposed to be ) but it is possible.
2011-10-19 08:53 PM
By definition MetroCluster is SyncMirror. To answer your question, you could run FMC with both nodes sharing the single shelf and only assign pool 0 disk. You would not mirror plexes. I believe normal CFO will function but DR takeover (forcetakeover -d) will not (since you do not have mirrored plexes).
As everyone mentioned, doing this is not recommended and probably not supported. Perhaps this would be configured for a lab where hardware is limited but definitely not in a production environment. I do not recommend this configuration.