This year, while creating the disaster backup server, I decided to but all 6 systems on one V490 with a 3310 SCSI array made into a ZFS filesystem with double parity RAIDz2 and a hot spare. I allocated 90% and then carved it up among the Oracle Application Servers and Oracle database.
A couple of things I wanted to make note of: I found that if you simply specify the mask in the zonecfg, you do not have to add any /etc/netmasks or /etc/network entries. After issuing an "add net" simply say "set address=192.168.1.1/24" and it saves times. It will, of course, assume the class C mask (which is correct in this case) but the ugly nag message can be avoided if you do either of the above.
The other thing I got bit by was sub-zone Ethernet connections. I had a sub-zone that needed a special VLAN not used by the global zone. So, I just added the Ethernet connection and forgot about it. After rebooting, NOTHING WORKED! It turns out the special VLAN was the connection used by my (virtual) DNS server. Since the global instance wasn't using it, it didn't get plumbed up after reboot and then it wasn't there for the DNS system and so nothing had DNS which caused a lot of slow timeouts, black screens, etc. I didn't want or need that VLAN in the global zone so I experimented with touching /etc/hostname.ce2 (the unused card in the global zone). That is, a created an empty file placeholder. Amazingly - it worked.
Thursday, March 11, 2010
Subscribe to:
Posts (Atom)