Oracle 11g R2 (GI) Startup Process

Oracle 11g R2 (GI) Startup Process

11gR2 cluster management software (GI) startup sequence and diagnostic methods

In this article, we will introduce the Starting sequence of 11gR2GI and introduce common problems and corresponding solutions during GI startup.

Basically, we can divide the GI startup process into three phases: ohasd, cluster building, and resource startup.

First, the ohasd stage.

1. scripts in the/etc/inittab File

H1: 35: respawn:/etc/init. d/init. ohasd run

>/Dev/null 2> & 1 </dev/null

Called to generate the following process

Root 4865 1 0 Dec02? 00:01:01/bin/sh/etc/init. d/init. ohasd

Run

So if you haven't found this process, it means

+ The init. ohasd script may not be called.

+ The OS is running at an incorrect level.

+ Some S * ohasd scripts are suspended, such as S96ohasd

+ Crsctl enable crs is not configured for GI)

Then, the ohasd. bin process will be started, and the OLR will be accessed at this time. Therefore, if ohasd. bin cannot work normally, you need to check whether the OLR exists and can be accessed normally. OLR is stored in $ GRID_HOME/cdata/$ {HOSTNAME}. olr

2. The ohasd. bin process starts the corresponding agents (orarootagent, oraagent, cssdagnet, and cssdmonitor) to start the initialization resources of the cluster. If you find that these agent processes cannot be started, they are often caused by problems with executable files in the path $ GRID_HOME/bin. For example, if there is a problem with file permission settings, the file resume uption.

Next, build the cluster.

1. The Mdnsd process discovers the node and all Nic information in the cluster through Multicast. Therefore, make sure that the NIC in the cluster supports multicast. In addition, the communication between nodes is normal.

2. Start the Gpnpd process, publish the bootstrap information required to build the cluster, and synchronize the gpnp profile among all nodes in the cluster. Of course, synchronization is implemented through mdnsd. Therefore, if there is a problem with this process, you need to confirm that the communication between nodes is normal, and gpnp

Profile (<gi_home>/gpnp/profiles/peer/profile. xml) exists and can be accessed.

3. The Gipcd process starts. This process is responsible for managing all the private networks (clusters) in the cluster.

Interconnect) Nic. Of course, the private network information is obtained through gpnpd. Therefore, if there is a problem with this process, you need to confirm that the gpnpd process is running properly.

4. The Ocssd. bin process is started. This process first discovers the Voting Disk through the information in the gpnp profile, and then obtains the private network information of the cluster through the gpnpd process, and establishes a connection with other nodes. Therefore, if ocssd. bin cannot run properly, you need to confirm the information

+ Gpnp profile exists and can be accessed.

+ The gpnpd process runs normally.

+ The asm disk or device of the voting disk can be normally accessed.

+ The Intranet Communication between nodes is normal.

5. Start other initialization processes: ora. ctssd, ora. asm, ora. cluster_interconnect.haip, ora. crf, ora. crsd

.

Note: The above process is performed at the same time. That is to say, ocssd. bin, gpnpd. bin and gipcd. bin starts at the same time until gpnpd. bin runs normally, ocssd. bin and gipcd. bin to obtain the corresponding information, in gpnpd. before bin is running properly, ocssd. bin and gipcd. the failure to access gpnp profile in bin can be ignored.

Finally, the resource is started. At this stage, various resources are started through the crsd process.

1. Start the Crsd process. This process needs to access OCR. If your OCR is stored on ASM, make sure that

The ASM instance runs normally and the ASM disk where the OCR is located has been loaded. If OCR is stored on a bare device, ensure that the corresponding device runs properly.

2. Crsd starts the corresponding agents (orarootagent,

Oraagent _ <rdbms_owner>, oraagent _ <gi_owner> ). If the agent cannot be started, it is often caused by a problem with the executable file in the path $ GRID_HOME/bin. For example, if the file permission settings are incorrect, the file upload uption.

3. Start all resources.

Ora. net1.network: network resource. This resource is used to manage the public network, scanvip, and vip of the cluster,

All listener resources depend on this resource. Therefore, if there is a problem with this resource, vip, scanvip, and listener will both be offline, you need to check whether there is a problem with the public network.

Ora. <scan_name>. vip: the vip resource corresponding to scan. A maximum of three vip resources can be created.

Ora. <node_name>. vip: the vip resource corresponding to the node.

Ora. <listener_name>. lsnr: Listener resource. Note that the listener. ora file is automatically generated starting from 11gR2 and does not need to be manually modified.

Ora. LISTENER_SCAN <n>. lsnr: scan listener.

Ora. <disk group name>. dg: ASM disk group resource. This resource is created when the disk group is mounted and deleted when the disk is dismounted.

Ora. <Database Name>. db: database resource. In 11gR2, instance resources no longer exist. New database resources manage all instances of the rac database, and which instances are included in the database, it is determined by the resource parameter "USR_ORA_INST_NAME @ SERVERNAME (<node name>. In addition, if your database is stored on the ASM disk, the database resources will depend on the corresponding disk group resources. This dependency is automatically added. However, if the database is transferred to another disk group, the original dependancy will not be automatically deleted. You need to manually delete the database (crsctl modify res ......).

Ora. <service name>. svc: Database Service resource. Starting from 11gR2, this resource has only one and will not be like 10gR2. Each database service resource contains two resources: srv and cs.

Ora. cvu: This resource was introduced from 11.2.0.2. It regularly performs cluvfy operations on the cluster to verify whether the cluster has some configuration problems.

Ora. ons: ONS resources, which are basically the same as those of previous versions.

In addition, we will briefly introduce the files to be viewed to diagnose GI startup problems.

$ GRID_HOME/log/<node_name>/ocssd <=

Ocssd. bin log

$ GRID_HOME/log/<node_name>/gpnpd <= gpnpd. bin log

$ GRID_HOME/log/<node_name>/gipcd <= gipcd. bin log

$ GRID_HOME/log/<node_name>/agent/crsd <= crsd. bin log

$ GRID_HOME/log/<node_name>/agent/ohasd <= ohasd. bin

Logs

$ GRID_HOME/log/<node_name>/mdnsd <= mdnsd. bin log

$ GRID_HOME/log/<node_name>/client <= logs of operations performed on clusters using the GI tool (ocrdump, crsctl, ocrcheck, gpnptool, and so on.

$ GRID_HOME/log/<node_name>/ctssd <= ctssd. bin

Logs

$ GRID_HOME/log/<node_name>/crsd <= crsd. bin log

$ GRID_HOME/log/<node_name>/cvu <= cluvfy log output.

$ GRID_HOME/bin/diagcollection. sh <= use this script to obtain more comprehensive diagnostic logs.

Finally, the cluster socket file (/var/tmp/. Oracle

Or/tmp/. oracle), because many processes in the cluster communicate through ipc, these socket files must exist and have the correct permissions.

In the above section, we will briefly introduce the GI startup sequence and basic diagnostic methods, hoping to help you diagnose GI startup problems.

Oracle 11g installation manual on RedHat Linux 5.8 _ x64 Platform

Installing Oracle 12C in Linux-6-64

Install Oracle 11gR2 (x64) in CentOS 6.4)

Steps for installing Oracle 11gR2 in vmwarevm

Install Oracle 11g XE R2 In Debian