Metalink -3rd party culsterware 

* Using Oracle Clusterware with Vendor Clusterware FAQ [ID 332257.1] 
* How To Integrate Symantec SFRAC with Oracle CRS and Troubleshoot the Results [ID 782148.1] 
* Oracle's Policy for Supporting RAC 10g with Symantec SFRAC on Unix [ID 397460.1] 
* Veritas ODM library no longer used after RDBMS patchset installation [ID 741015.1] 
* 10g RAC: Steps To Increase CSS Misscount, Reboottime and Disktimeout [ID 284752.1] 
 

notes
-----

- RAC 10g 10.1 > party vendor clusterware (any vendor) cannot replace CRS and must always be integrated with CRS (a CRS demoni jsou videt v v /etc/inittab)
- With 3rd party clusterware installed, the oprocd process should NEVER be running


ORACLE má API pro
-----------------
* IPC - VCSIPC  - musi se objevit v alert logu pri startu instance (IPC: Veritas)
      - for Global Enqueue Service locking traffic and Global Cache Service cache fusion., System Kernel Generic Interface Inter-Process Communications (skgxp), that isolates Oracle from the underlying transport
      installing and configuring SF Oracle RAC - links Oracle with the Veritas IPC library (to znamena ze veritas muze sledovat, co si RAC přes interconnect posila)
      knihovna /opt/ORCLcluster/lib/libskgxn2.so (cluster membership monitor library)

* IO data (ODM) - musi se objevit v alert logu pri startu instance (ODM: Veritas) 
   - Diskove subsystemy, volume managery , cluster FS a další ptakoviny


Veritas Storage Foundation™ for Oracle RAC (SFRAC)
==========================================

http://sfdoccentral.symantec.com/Storage_Foundation_HA_50MP2_HPUX.html

clusterware a volume managment (volume manager + cluster FS)

- ma vlastni data IO stack
- database I/O API definovane Oraclem (implementace ODM - Veritasi library)
- clusterware ma vlastni protokoly 
  - pro interconnect (LLT [low latency transport] - heartbeaty, load balance) - predelavka IP stacku, niz uz je jen linkova vrstva (ethernetove ramce nebo jine)
  - cluster membership - Group Membership Services/Atomic Broadcast (GAB)

- cluster sluzba je zde - VCS - Veritas cluster server

 

VCS
===

VCS uz pracuje s komponenty/resource, monitoronig, notifikace

daemoni:
--------
- High Availability Daemon (HAD) – definice resource group a services group, dependences / cluster management
- každý HAD ma nad sebou lokalni HAD agenty, kteří si povidaji s ostatnimi na clusteru přes GAB/LLT, jednotlive udalosti lze logovat
 HAD waits in a loop waiting for messages from agents, ha commands, the graphical user interfaces, and the other systems.

 
HAD Agenti
----------
Each agent process has two components: the agent framework and the agent entry
points. The agent framework provides common functionality, such as
communication with the HAD, multithreading for multiple resources,
scheduling threads, and invoking entry points. Agent entry points implement
agent-specific functionality. Follow the performance guidelines below when

configuring agents.

Agent pro Oracle (Single instance?)
http://sfdoccentral.symantec.com/sf/5.0/hpux/pdf/vcs_oracle_install.pdf (je zde popis jak se mají spravne nastavit service groupy)
·         Agenti jsou dva (jeden pro oracle dabazi, druhy pro listener)

 
Instalace SF pro Oracle RAC
http://sfdoccentral.symantec.com/sf/5.0/hpux/pdf/sfrac_install.pdf
 

■ Veritas Cluster Server (VCS) - VCS is configured to enable agents to direct and manage the resources required by Oracle RAC (via SRVCTL or othr API?)

■ Veritas Volume Manager (VxVM)

■ Veritas File System (VxFS)

■ Veritas High Availability Agent for Oracle

■ Veritas Volume Replicator (VVR)

■ Other SF Oracle RAC modules

 
VCS is configured to enable agents to direct and manage the resources required 
by Oracle RAC (via SRVCTL or othr API?)