Dear All,
In this post i have explained the step-by-step oracle 19C RAC startup sequence.
The startup
sequence in Oracle 19c RAC is complex due to the multi-instance, multi-node
nature of RAC.
1. Start Oracle Grid Infrastructure
(Clusterware) – crsctl
2. Start Oracle ASM (Automatic
Storage Management) - srvctl / as part
of cluster start
3. Start Oracle RAC Database
Instances using srvctl
4. Start Oracle Net Listener
5. Start Oracle Services (optional)
6. Start ASM (if not already done)
This
process ensures that all instances across nodes are properly synchronized,
ensuring high availability and data consistency in a shared-disk
environment like Oracle RAC. Proper startup ensures that all Oracle processes,
database services, and storage are correctly initialized, providing a stable
environment for your applications and database users.
Oracle 19c RAC Startup Sequence: Detailed
Explanation
In Oracle
19c Real Application Clusters (RAC), the startup sequence refers to
the steps Oracle follows to bring the database and its associated services into
an operational state. The startup process is different in a RAC environment
compared to a single-instance Oracle database because it involves starting and
coordinating multiple instances across several nodes, ensuring they can work
together in a shared-disk environment.
The startup
sequence in Oracle RAC is divided into the following stages:
1. Starting Oracle Grid Infrastructure (CRS) and Cluster ware
Before
any Oracle RAC database instance starts, Oracle Cluster ware (Grid
Infrastructure) must be started on all nodes in the cluster.
1: Start the Cluster ware Stack (CRS)
> Oracle Cluster ware manages the nodes in the
cluster and ensures they function as a single logical unit. This includes
monitoring the health of all instances, providing failover mechanisms, and
managing resources like voting disks and OCR (Oracle Cluster
Registry).
> Cluster ware services (CRS)
must be running on each node to manage the cluster's health.
Commands:
$ crsctl start crs # Start Oracle Clusterware (CRS) on all
nodes
When you
issue this command, the following services are started:
Clusterware Daemons: These include crsd, evmd, and cssd.
crsd (Cluster Ready Services Daemon): Manages Oracle Clusterware resources.
evmd (Event Manager Daemon): Collects and processes events in the cluster
cssd (Cluster Synchronization Services Daemon): Manages cluster membership and coordinates node status and interconnect activity.
Start Oracle Clusterware (Oracle ASM)
> If you're using Oracle
ASM (Automatic Storage Management) for your RAC database, it must also
be started as part of the Clusterware stack.
> ASM is responsible for managing
the storage for Oracle databases in a RAC environment. ASM organizes data
into disk groups and provides striping and mirroring for high
availability.
Commands:
$ crsctl start asm
This
command starts the ASM instance, and it will mount the ASM disk groups
that contain the Oracle database files, including voting disks
and OCR (Oracle Cluster Registry).
Starting
Oracle RAC Database Instances
Once Clusterware
is up and running, the next step is to start the Oracle RAC database
instances. This step starts each instance of the RAC database on each node
in the cluster.
>> Start the Database Instances Using srvctl
To start
a database in Oracle RAC, you typically use the srvctl command. This command ensures that all instances
of the RAC database are properly started, ensuring correct inter-node
communication and Cache Fusion functionality.
> srvctl automatically starts all
required Oracle background processes, including LGWR (Log Writer), DBWn
(Database Writer), and PMON (Process Monitor), and ensures proper
interconnect activity between nodes.
Commands:
$ srvctl start database -d <dbname>
ex: <dbname> is the name of the RAC
database that you want to start.
$srvctl start database -d racsdb
2. Oracle Instance Startup Sequence
Once the srvctl command is issued, Oracle begins the following
instance startup sequence for each node in the cluster:
Process Initialization:
> The instance processes like SMON,
PMON, and others are started. These processes ensure the instance runs
in a healthy state.
Instance Registration with the Listener:
> Each instance registers itself
with the Oracle Net Listener for client connections. It ensures that the
listener knows which instance is available to service requests for that
particular instance.
Shared Memory Allocation:
>SGA (System Global Area) and PGA (Program Global Area)
are allocated for each instance , SGA is a shared memory region that
contains data and control information for the database.
> PGA is a private memory region used
by Oracle background processes and user sessions.
Global Enqueue Initialization:
Oracle RAC uses a shared global
cache to manage locks and latches on data blocks across all nodes. The
instances communicate and synchronize via Global Enqueue Services (GES)
and Global Cache Services (GCS).
> The Global Cache layer
ensures consistency of data across RAC nodes.
Database Opening: After the instance processes and
memory structures are initialized, the Database itself is opened. Oracle
opens the database in READ ONLY mode to ensure that it can be accessed
by all instances of the RAC database, and parallel execution is
available across nodes.
>The redo log files are
opened to allow transaction processing.
Cache Fusion Initialization:
> Cache Fusion ensures that data blocks are
shared across instances. Oracle will initiate Global Cache Synchronization
to synchronize the shared memory structures and data blocks between instances.
> This is key in Oracle RAC to
ensure all nodes are consistent with the data they share.
Database Processes Start: Database processes, such as LGWR
(Log Writer), DBWn (Database Writer), SMON (System Monitor), and PMON
(Process Monitor), are started. These processes are responsible for maintaining
the health of the database and managing transactional and background
operations.
Verify
Database Instance Startup
You can
verify the status of your RAC database instances and ensure that they are up
and running with the following command:
$ srvctl status database -d <dbname>
The
output will indicate which instances are running and their status across all
RAC nodes.
3. Starting Oracle Net Listener
The Oracle
Net Listener is a process that listens for incoming client connection
requests and directs those requests to the appropriate database instance in the
cluster.
Start Oracle Listener
Use the srvctl utility to start the listener for Oracle RAC:
Commands:
$ srvctl start listener -l <listener name>
This
command starts the listener for your Oracle RAC database, allowing client
connections to be established.
4. Start Oracle Services -- Optional
Oracle
services allow you to manage the load balancing and high availability of the
database in a RAC environment. For example, Oracle can use Service Names
to direct clients to the appropriate instance based on load or availability.
Start Oracle
Services Using srvctl
$ srvctl start service -d <dbname> -s
<service_name>
This
command starts the specified service on the available instances. In a RAC
setup, Oracle can configure services to run on specific nodes or across all
nodes in the cluster.
Verify Service Status
To check
the status of the services, you can use the following command:
$ srvctl status service -d <dbname> -s
<service_name>
5. Starting Oracle Automatic Storage Management
(ASM)
ASM
(Automatic Storage Management) is responsible for managing database storage in
RAC environments. It needs to be started to manage disk groups and handle file
operations.
Step 1: Start ASM Using srvctl
You can
start ASM using the srvctl command
as follows:
$ srvctl start/status/stop asm
6. Database and Instance Synchronization in RAC
Once all
instances are up, and the Oracle Grid Infrastructure is running, Oracle
RAC ensures that all instances are synchronized via Cache Fusion. This
synchronization allows multiple instances to access the same set of database
files (e.g., data files, control files, and redo log files) in a shared storage
configuration.
Monitor Cluster Synchronization
You can
monitor Cache Fusion performance using the v$gc_stats view:
crsctl
check ctss
crsctl
check css
crsctl
check cluster
ways to
check cache fusion in Oracle RAC:
GCS
statistics: GCS
statistics for current and cr blocks, such as gc current blocks received and gc
cr blocks received, can be used to analyze the effect of cache fusion.
GCS wait
events: GCS
wait events, such as gc current block 3-way and gc cr grant
2-way, can be used to analyze the effect of cache fusion.
V$SYSSTAT
view: The V$SYSSTAT view can be queried to monitor global cache statistics.
Cache Fusion is a feature in
Oracle RAC that allows multiple instances to share and read the same data block
concurrently. It uses the Global Cache Service (GCS) to check if another
instance has a copy of a data block before accessing it. If another
instance has the block, Cache Fusion transfers it directly between the
instances without writing it to disk
Thanks,
Srini
