In this short article, we will cover in brief the main changes for the CRS part of Oracle Grid Infrastructure (GI) in 18.104.22.168. Remember that 22.214.171.124 is a patchset release and is available only to Oracle customers with a support contract that provides access to patches.
Software Update Before Install or Upgrade A new installer screen, as shown in Figure 1, will allow you to download software updates that include installation prerequisite information, bug fixes, and Oracle patches from My Oracle Support. These will be installed as part of the installation or upgrade process. This removes the old requirement of installing and then applying any necessary patches.
Automatic NIC Bonding for Private Networks 126.96.36.199 introduces the ability for the clusterware to automatically bond private interconnect interfaces. In previous releases, this has be done using third-party/OS software. If you have multiple NICs that can be used for private interconnects, you can simply mark them all as private on the network selection screen in the OUI or Configuration Wizard. Up to four NICs will be used at any time for load balancing and failover, and interfaces can be added or removed using the oifcfg command. You will see that a VIP is started on each of the active interfaces and this VIP is relocated if any failure occurs.
Cluster Health Monitor The cluster health monitor was previously downloaded from Oracle Technology Network (OTN) and monitored load on the system specifically to help when a hang or system issue would cause a cluster or node outage. This will now be delivered with GI, though the Berkeley Database it uses to store data is to be used only for the monitor. You would have to license Berkeley separately to use it for your applications. OHASD manages a new resource to ensure that the monitor is running and storing the system statistics.
Grid Installation Owner Group Requirements The grid installation owner no longer has to be a member of the group designated for ASMOPER if one has been created.
Out-of-place Install for Patchset Releases The patchset release is now a complete release that can be installed as a new installation or a patch. The software will be installed out of place (in its own separate home) and can be patched easily prior to being applied. This reduces the downtime required to apply a patchset on each node and allows patches to be applied to the patchset before upgrade. The following steps are required to install this as a patchset:
1. Provision local disk space on all cluster nodes for the new GI home (5.5GB at least on Linux). This is on top of the space for the old home.
2. If upgrading from 188.8.131.52, use ASM in a rolling manner; then patches for bugs 9413827 and 9706490 need to be applied to the existing GI home. See MOS Note 1274629.1. For other starting versions, check the README for any prerequisite patches.
3. If you are using udev managed devices, make sure that all the parameter names are in uppercase and the following parameter order is used:
KERNEL==”xvdb1″, OWNER=”grid”, GROUP=”asmadmin”, MODE=”660″
4. Run the OUI and ensure that the upgrade option is chosen.
5. Run rootupgrade.sh when prompted to do so.
Clusterware Configuration Wizard 184.108.40.206 includes a clusterware Configuration Wizard that can be used to configure the cluster after a software-only installation or upgrade. It effectively runs the configuration part of the OUI, which saves you having to edit parameter files before running root.sh, as was previously required.
Automatic CVU Checks The CVU will run checks at specified intervals to report any problems it sees in the cluster setup. This is a resource managed by CRSD, and you can see/change the check frequency using srvctl:
[grid]$ /u01/app/11.2.0/grid/bin/srvctl config cvu CVU is configured to run once every 360 minutes
You can also see the CVU checks’ output reported to the clusterware alert log.