Showing posts with label rac. Show all posts
Showing posts with label rac. Show all posts

Tuesday, 6 June 2017

What are Nodeapps Services in Oracle RAC


Nodeapps are standard set of oracle application services which are started automatically for RAC.

Node apps Include:

1) VIP.
2) Oracle Net listener.
3) Global Service Daemon.
4) Oracle Notification Service.

Nodeapp Services run on each node of the cluster and will switched over to other nodes through VIP during the failover.


Useful commands to maintain nodeapps services:

srvctl stop nodeapps -n NODE1 
[ STOP NODEAPPS  on NODE 1 ]
srvctl stop nodeapps -n NODE2            
[ STOP NODEAPPS  on NODE 2 ]

srvctl start nodeapps -n NODE1              
[ START NODEAPPS on NODE1  ]
srvctl start nodeapps -n NODE2             
[  START NODEAPPS ON NODE2 ]

srvctl status nodeapps                      
[Check the status of services on all nodes]

Friday, 5 May 2017

Oracle Database Administrator Interview Questions on RAC

Oracle RAC (Voting Disk ) FAQ's

1.What is Voting Disk?

Voting Disk is a file that maintains node membership details.
All members in the Cluster read and write Heartbeat Information in Voting Disk.

2. What Information is stored in Voting Disk?

Voting disks contain static and dynamic data.
Static data : Info about nodes in the cluster
Dynamic data : Disk heartbeat logging
It maintains and consists of important details about the cluster nodes membership, such as
- which node is part of the cluster,
- which node is joining the cluster, and
- which node is leaving the cluster.

3. Why do we need Voting Disk?

CSSD processes (Cluster Services Synchronization Daemon) monitor the health of  RAC nodes employing two distinct heart 

beats: Network heart beat and Disk heart beat. Healthy nodes will have continuous network and disk heartbeats exchanged 

between the  nodes. Break in heart beat indicates a possible error scenario. There are few different scenarios possible 

with missing heart beats:
1. Network heart beat is successful, but disk heart beat is missed.
2. Disk heart beat is successful, but network heart beat is missed.
3. Both heart beats failed.
In addition, with numerous nodes, there are other possible scenarios too. Few possible scenarios:
1. Nodes have split in to N sets of nodes, communicating within the set, but not with members in other set.
2. Just one node is unhealthy.
Nodes with quorum will maintain active membership of the cluster and other node(s) will be fenced/rebooted.

4.Why do we have odd number of voting disks?

Odd number of Voting disks are required to prevent Split Brain Syndrome.
A node must be able to access more than half of the voting disks in order to decide which node can be evicted incase of 

failure.

5. How to backup Voting Disk?

Prior to Oracle 11g R2 version, Voting Disk is backed up using dd command

From 11g R2, no need to manually backup voting disk.
It is automatically backup along with OCR  whenever there is a configuration change.

In 11g R2, Restoring Voting disk from a manually copied backupfile may prevent cluster services from starting up.

Friday, 30 December 2016

What is OCR (Oracle Cluster Registry)?

Oracle Cluster Registry

OCR is the central repository for the Oracle cluster.
OCR has information about Port information,node failures, node reconfiguration,Database Status,Listener Status,Instance status, Network Status in real time.


Default Location of ocr.loc file on linux --  /etc/oracle/ocr.loc

Oracle Clusterware reads ocr.loc to determine the registry information and application resources that need to started up on each cluster node.

The below processes update OCR with realtime information.

1. CRSd updates  OCR with  information about node failure 
2. CSSd updates the OCR when a node is added or deleted
3. NetCA, DBCA, SRVCTL updates  OCR with  services related information

OCR also has information about  Software active version

Software  version can also be found using below command
$ crsctl query crs activeversion



OCR Backup and Restore



OCR is a binary file and cannot be edited manually.By Default, OCR is backed up automatically every 4 hours (from Instance Startup time).
OCR backups are cached across all nodes of a cluster.However,Only Master RAC node can write to OCR.
OCR backups are overwritten automatically. Clusterware maintains last 3 backups of OCR. 

Backup location can be defined using the below command.

$ocrconfig -backuploc <backup location>

It is recommended to use shared location for storing OCR backups.

To view the current OCR backups available, use the below commands

$ocrconfig –showbackup auto 

$ocrconfig –showbackup manual


After any clusterware related changes, OCR backup can be taken manually.

$ocrconfig –manualbackup


If OCR file is Corrupted or lost, CRS Instance may crash.

To restore OCR from Backup,follow the procedure below


1.Using ocrconfig identify the latest OCR backup using command below

$ocrconfig -showbackup
2.Shutdown CRS Services on all RAC nodes
$crsctl stop crs
  
3.Start CRS on one node in exclusive mode
$crsctl start crs -excl 
Stop crsd , if it is running

$ crsctl stop resource ora.crsd -init

4.Restore the OCR 
$ocrconfig –restore <backuplocation/backupfilename>
   
5. Verify the integrity of OCR
$ocrcheck
6. Shutdown  CRS Services on the node where you had started in exclusive mode
$crsctl stop crs 


7.Start CRS on all RAC nodes 
$crsctl start crs
8. Verify the restored OCR 
$cluvfy comp ocr –n all -verbose

Thursday, 17 March 2016

How to Perform Clusterware Compatibility Testing in Oracle RAC

+ Compatiblity Testing can be done using Oracle Certification Environment (OCE) kit

Installing the Oracle Certification Environment Software for Oracle RAC

The OCE Certification Kit required to certify the system for Oracle RAC 11g Release 1 (11.1) is available for download only. The Single Instance certification tests should be completed prior to installing the OCE kit for Oracle RAC. Refer to the previous section if necessary. Once Single Instance testing has successfully completed, the single instance OCE installations must be archived to allow OCE installations for Oracle RAC to succeed. The OCE kit for Oracle RAC should be installed separately on each node of the cluster. If the ORACLE_HOME is located on a shared disk, multiple installations of OCE will not be possible. In that case, it will not be possible to run OCE tests simultaneously, and the time required to complete certification will be greatly increased. To install the OCE Kit:
1.      Download OCE for Oracle RAC 11g Release 1 (11.1) archive to a suitable location, such as /tmp/oce. The OCE archives are either CPIO archives, or compressed CPIO archives.
If compressed, extract as follows:
gunzip -c OCE ARCHIVE | cpio -idmv
If not compressed, extract as follows:
cpio -idmv < OCE_ARCHIVE
Where, OCE_ARCHIVE is the name of the archive.
2.      Enter the following command to run the OCE Installer:
3.  $ archive_location/oce_install.sh
The Environment variable screen is displayed.
4.      Enter values for each of the environment variables as described on screen.
Note: You must press the Enter key once to enter the new variable value, and then press the Enter key again to move onto the next variable.
5.      Type Done when finished and press the Enter key.
The installation progress screen is displayed. When all stages are complete, the installer will exit.
6.      Check $ORACLE_HOME/OCEinstallRAC.log file, and verify that there are no errors.
7.      Download the Binaries Package for OCE for Oracle RAC 11g Release 1 (11.1) which matches your platform, and extract the archive as explained in Step 1.
8.      Enter the following command to run the OCE binaries installation script.
9.  $ /tmp/oce/oce_exes_install.sh
10.  Check the OCE Kit installation log file, $ORACLE_HOME/OCE/install_log.txt, and verify that the installation was successful.
The kit is installed in the $ORACLE_HOME/oce directory.

Preparing the System for Multi-Node High Availability Services Testing
The Oracle RAC High Availability Services test suite must use shared storage for data files. Depending on the type of shared storage utilized, some preliminary setup may be required. Before running the Oracle RAC High Availability Services test suite, complete the following:
  • If you are using raw devices or logical volumes for shared storage perform the following steps:

Note:
In this example, the OCE user is oracle, which is a member of the dba group; there are 4 nodes in the cluster; and the OCE logical volumes are located in /dev/ocevg/ directory.

    1. Set up the devices or logical volumesrequired by the tests.
    2. Ensure raw devices are accessible and writable across all nodes by the OCE user
      1. # chown -R oracle:dba /dev/ocevg
      2. # chmod -R og+w /dev/ocevg
    3. Export $ORACLE_HOME/oce/work$ORACLE_HOME/dbs, and $ORACLE_HOME/network/admin from the node 1 to all other nodes in the cluster.
On node 1:
      1. # exportfs -i -o rw <node2>:$ORACLE_HOME/oce/work \
<node3>:$ORACLE_HOME/oce/work \
<node4>:$ORACLE_HOME/oce/work
      1. # exportfs -i -o rw <node2>:$ORACLE_HOME/dbs \
<node3>:$ORACLE_HOME/dbs \
<node4>:$ORACLE_HOME/dbs
    1. # exportfs -i -o rw <node2>:$ORACLE_HOME/network/admin \
<node3>:$ORACLE_HOME/network/admin \
<node4>:$ORACLE_HOME/network/admin
    1. $ORACLE_HOME/oce/work$ORACLE_HOME/network/admin, and $ORACLE_HOME/dbs must be mounted on all secondary nodes from the primary (exported) node.
On all nodes except node 1:
      1. # mkdir –p $ORACLE_HOME/oce/work
      2. # chown oracle:dba $ORACLE_HOME/oce/work
      3. # mount <node1>:$ORACLE_HOME/oce/work $ORACLE_HOME/oce/work
      4. # mount <node1>:$ORACLE_HOME/dbs $ORACLE_HOME/dbs
      5. # mount <node1>:$ORACLE_HOME/network/admin \
$ORACLE_HOME/network/admin
  • If you are using OCFS or NAS or a vendor clustered file system (CFS), and ORACLE_HOME directory is not located on shared partition, then perform the following steps:

Note:
If NAS, ensure that the appropriate mount options are employed when mounting the NAS partition. Oracle requires specific mount options. Consult your NAS Filer documentation for further details.

    1. Symbolically link $ORACLE_HOME/dbs to the OCFS/CFS/NAS partition on all nodes (in this example, the OCFS/CFS/NAS partition is at /sharedfs).
On node 1:
    1. mkdir /sharedfs/dbs
    2. chown oracle:dba /sharedfs
On all nodes:
      1. mv $ORACLE_HOME/dbs $ORACLE_HOME/dbs.BAK
      2. ln -s /sharedfs/dbs $ORACLE_HOME/dbs
    1. Export $ORACLE_HOME/oce/work and $ORACLE_HOME/network/admin from the primary node.
      1. # exportfs -i -o rw <node2>:$ORACLE_HOME/oce/work \
<node3>:$ORACLE_HOME/oce/work \
<node4>:$ORACLE_HOME/oce/work
    1. # exportfs -i -o rw <node2>:$ORACLE_HOME/network/admin \
<node3>:$ORACLE_HOME/network/admin \
<node4>:$ORACLE_HOME/network/admin
    1. $ORACLE_HOME/oce/work and $ORACLE_HOME/network/admin must be mounted on all secondary nodes from the primary (exported) node. Default mount options will suffice.
On all nodes except node 1:
      1. # mkdir –p $ORACLE_HOME/oce/work
      2. # chown oracle:dba $ORACLE_HOME/oce/work
      3. # mount <node1>:$ORACLE_HOME/oce/work $ORACLE_HOME/oce/work
      4. # mount <node1>:$ORACLE_HOME/dbs $ORACLE_HOME/dbs
      5. # mount <node1>:$ORACLE_HOME/network/admin \
$ORACLE_HOME/network/admin
  • If you are using OCFS or CFS accessing a shared Oracle home directory, no setup is required.
  • Ensure that no databases are running.

Starting Test Manager
To start Test Manager:
  1. Ensure that the DISPLAY environment variable is set appropriately for your system. To verify that it is, try starting up xclock. If you do not see the clock, or you receive errors, DISPLAY is not set appropriately. You must correct any errors before proceeding.
  2. Enter the following command to launch OCE Test Manager:
3.  $ORACLE_HOME/oce/bin/startTM.sh > /tmp/OCETM.log 2>&1
The OCE Main Menu and OCE Test Manager windows appear.
Running a Test for the First Time
If this is your first time running certification tests, you must perform the following steps:
  1. Start Test Manager as described in Starting Test Manager.
  2. From the OCE - Main Menu window, double click Utilities.
  3. Run the bmchk test by selecting it and clicking Execute.
  4. When the test completes, click Results in the Test Manager window to check the outcome. If the test fails, you must analyze the output ($INST_HOME/work/bmchk) and resolve any issues. Do not proceed with testing until bmchk executes successfully.
  5. Run sdbck (the Seed Database Verification utility) test by selecting it and clicking Execute.
  6. When the test completes, click Results in the Test Manager window to check the outcome. If the test fails, you must analyze the output ($INST_HOME/work/sdbck) and resolve any issues. Do not proceed with testing until sdbck executes successfully.
  7. Run cssck (the CSS Daemon Verification utility) test by selecting it and clicking Execute.
  8. When the test completes, click Results in the Test Manager window to check the outcome. If the test fails, you must analyze the output ($INST_HOME/work/cssck) and resolve any issues. Do not proceed with testing until cssck executes successfully.

Running the OCE Test Suites

The OCE release consists of a set of test suites that you run from Test Manager. Each test suite consists of one or more individual tests. To complete the certification, run each of the Test Suites in the kit for the product for which you are certifying your system.
1.      From the OCE - Main Menu screen, double-click Complete test suites.
2.      From the screen that appears, select the test suite you want to run and click Execute to run it.
The test suite runs.
Test Manager creates two entries for the test suite in the Test Manager window; one in the Suite Name field and another in the History field:
o    The entry in the Current Test field is displayed only for the duration of a test. It displays the time at which you requested the test, and if it starts, when it started. Test Manager might display some tests with a status of Waiting until resources become available on the system.
o    The entry in the History field displays the time you requested it.
When a test finishes, Test Manager deletes its entry in the Current Tests field and adds another entry to the History field showing when the test finished.






Below are the Test plans for Oracle Clusterware Compatibility (Destructive) Testing
(Category : ORACLE HIGH AVAILABILITY FEATURES)
Clusterware Test Category
[Test Code]
Action  Target
Detailed Test Execution
Expected Test Outcome
Actual Test Outcome
[D]
Oracle HA Features


[HW-CW-09]

Run multiple cluvfy operations during Oracle Clusterware and RAC install  All RAC hosts

Configuration:
GNS:
Gns with dhcp (1)
Gns without dhcp (2)
Without gns (3)
Preferred option 1, if not applicable option   2,  if still not applicable option 3

ASM:
Flex asm (1)
Standard asm (2)
Preferred option 1, if not applicable option 2

DB:
CDB
Preconditions:
·          Type `cluvfy` to see all available command syntax and options

Steps:
1- Run cluvfy precondition
2- Do the next install step
3- Run cluvfy post-condition
(cluvfy comp software –n node_list) to check the file permissions
No need to collect CRS/RDBMS log for this test.  You need to submit the output for cluvfy.


Vendor Clusterware:
- same as RAC

RAC:
-           Correct cluster verification checks given the state of the cluster hardware and software

Pls provide cvu related logs under
$CRS_HOME/cv/log


[HW-CW-10]

Run concurrent crsctl start/stop crs commands to stop or start Oracle Clusterware in planned mode  All RAC hosts

Configuration:
GNS:
Gns with dhcp (1)
Gns without dhcp (2)
Without gns (3)
Preferred option 1, if not applicable option 2,  if still not applicable option 3

ASM:
Flex asm (1)
Standard asm (2)
Preferred option 1, if not applicable option 2

DB:
CDB


Preconditions:
·          Initiate all Workloads
·          Identify both CSS and CRS master nodes
·          Type `crsctl` as root to see all available command syntax and options

Steps:
1- As root user, run `crsctl stop crs` command concurrently on more than one RAC host, to stop the resident Oracle Clusterware stack
2- Wait until the target Oracle Clusterware stack is fully stopped (via `ps` command)
3- As root user, run `crsctl start crs
    -wait` command concurrently on more than one RAC host, to start the resident Oracle Clusterware stack


Vendor Clusterware:
- N/A

RAC:
Stop:  All Oracle Clusterware daemons stop without leaving open ports or zombie processes
Start:  All Oracle Clusterware daemons start without error messages in stdout or any of the CRS, CSS or EVM traces
Start:  All registered HA resource states match the “target” states, as per 
“crsctl stat res –t”

For 12cR1, collect
“crsctl stat res –t” in a 60s loop from beginning till the end of run.  Attach the output for auditing.

[HW-CW-11]

Run other concurrent crsctl commands, such as crsctl check crs,   All RAC hosts

Configuration:
GNS:
Gns with dhcp (1)
Gns without dhcp (2)
Without gns (3)
Preferred option 1, if not applicable option   2,  if still not applicable option 3

ASM:
Flex asm (1)
Standard asm (2)
Preferred option 1, if not applicable option 2

DB:
CDB
Preconditions:
·          Initiate all Workloads
·          Identify both CSS and CRS master nodes
·          Type `crsctl` as root to see all available command syntax and options

Steps:
1-        As root user, run any `crsctl check crs` commands concurrently on all nodes
2-        As root user, run any `crsctl check cluster -all` commands concurrently on all nodes


Vendor Clusterware:
- same as RAC

RAC:
-           Both `crsctl check crs` and `crsctl check cluster -all` commands produce the appropriate, useful output, without any error messages
-           Collect output for step 1 and step 2

[HW-CW-12]

Votedisk and OCR operation

Configuration:
GNS:
Gns with dhcp (1)
Gns without dhcp (2)
Without gns (3)
Preferred option 1, if not applicable option 2,  if still not applicable option 3

ASM:
Flex asm (1)
Standard asm (2)
Preferred option 1, if not applicable option 2

DB:
CDB
Preconditions:
·          Make sure votedisk on ASM diskgroup
·          Make sure ASM OCR files are used
·          Make sure at least one normal redundancy ASM Diskgroup with three failgroups is created and its “compatible.asm” attribute is set to “11.2”;

Steps:
1-        Make sure crs stack are running in all nodes.

2-        Run “crsctl query css votedisk” to check configured VFs;
3-        Run “crsctl replace votedisk +{ASM_DG_NAME}”(As crs user or root user);
4-        Run “crsctl query css votedisk” to get the new VF list;
5-        Run “ocrconfig –add +{ASM_DGNAME}” as root user;
6-        Run “ocrcheck” to verify the OCR files;
7-        Restart CRS stack and then verify the VF/OCR after it comes back;

Variants:
1. Add up to 5 OCR files and restart CRS stack;


RAC:
-           In 12cR1, we can support up to 5 OCRs;

-            





[HW-CW-13]
crsctl command to manage Oracle clusterware stack

Configuration:
GNS:
Gns with dhcp (1)
Gns without dhcp (2)
Without gns (3)
Preferred option 1, if not applicable option 2,  if still not applicable option 3

ASM:
Flex asm (1)
Standard asm (2)
Preferred option 1, if not applicable option 2

DB:
CDB
Preconditions:
·          CRS stack is up and running on all nodes.

Steps:
1-        Run ‘crsctl check cluster –all’ to get the stack status on all cluster nodes. Make sure stack status of all cluster nodes are correct;
2-        Run ‘crsctl stop cluster –all’ to stop all CRS resource (CSSD/CRSD/EVMD) with application resources;
3-        Run ‘crsctl status cluster –all’ to make sure CRS resource are OFFLINE;
4-        Run ‘crsctl start cluster –all’ to bring back the whole cluster stack
RAC:
-           After running “crsctl stop cluster –all”, make sure all ocssd/evmd/crsd processes are stopped on all cluster nodes by “ps –ef”.
For 12cR1, collect
“crsctl stat res –t” in a 60s loop from beginning till the end of run.  Attach the output for auditing.


[HW-CW-14]

OCR stores in ASM’s diskgroup and kill asm fatal process

Configuration:
GNS:
Gns with dhcp (1)
Gns without dhcp (2)
Without gns (3)
Preferred option 1, if not applicable option   2,  if still not applicable option 3

ASM:
Flex asm (1)
Standard asm (2)
Preferred option 1, if not applicable option 2

DB:
CDB
Preconditions:
·          Initiate Workloads

Steps:
·          Make sure only ASM OCR files   are used by “ocrcheck –config”;

·          Kill the ASM pmon process on   the CRSD PE Master node;

Variants:
   Repeat the same test on non-OCR Master node.
Clusterware:
Because OCR is stored in ASM, if ASM fails or is brought down on crsd pe master, CRSD pe master will exit and  select a new crsd pe master

- ASM, CRSD will be automatically restarted.

-RDBMS instance should connect to other available asm instance in flex asm env 

- After CRSD restart, all resources’ state shouldn’t change

- New crsd pe master node should be the old crsd pe standby master A new crsd pe standby master should be   elected on other nodes.

(CRSD should recover resources’ previous state)

For 12cR1, collect
“crsctl stat res –t” in a 60s loop from beginning till the end of run.  Attach the output for auditing.



Collect Logfiles
Run each destructive test, taking note of the test start time, test stop time and fault injection time.  On the surviving node (if applicable), run the “date; crsctl stat res –t; sleep 60” in a loop

At the end of the test run, please collect the following logs and put them in directory <CRSHome>/log with the name format as [log_name]_[hostname] and then tar up and compress  with file name <VendorName>_<TestCode>.tar.gz.(e.g. WidgetCorp_HW-STOR-07.tar.gz):
·         Under <CRSHome>/log/<hostname>, the following logs are required
o   alert[hostname].log
o   crsd/crsd.log
o   cssd/ocssd.log
o   evmd/evmd.log
o   ohasd/ohasd.log
o   gpnpd/gpnpd.log
o   diskmon/diskmon.log
o   mdnsd/mdnsd.log
o   ctssd/ctssd.log
o   agent/*
o   gipcd/gipcd.log  (11.2.0.2 new feature)
o   cvu/cvulog/*.log (11.2.0.2 new feature)
o   cvu/cvutrc/*     (11.2.0.2 new feature)
o   srvm/*
o   admin/*
o   acfs/*
o   crfmond/*
o   crflogd/*
o   racg/*
o   gnsd/* (if gns configured)