Oracle fails if one of two disks fail

Hi Everyone,
I am new to Oracle RAC and have done a sample setup as per Jeffery Hunters instructions (http://www.oracle.com/technology/pub/articles/hunter_rac10gr2_iscsi.html).
Which by the way is an excellent document for us beginners.
When I built the test RAC so that I could do a proof of concept for my bosses I added a second separate disk (ISCSI setup - openfiler2).
With OCFS2 I setup and mounted 2 volumes. (U02 and U03)
ASM has the rest of the disks as (ASM1,VOL1 and ASM2,VOL2)
This way I have an OCRFile and CSSFile on each disk plus ASM mirroring the ORACLE datafiles on each drive. ( I think )
Now my problem is that if a fake a failure of one of my disks the nodes will reboot and I am unable to get the DATABASE back up.
Shouldn't the Cluster stay up and continue running on the one disk.
Any pointers would be appreciated.

Hi,
CRS is not starting up.
[root@linuxora1 bin]# ./crsctl check crs
Failure 1 contacting CSS daemon
Cannot communicate with CRS
Cannot communicate with EVM
[root@linuxora1 bin]#
I checked the /var/log/messages and see:
May 22 05:30:44 linuxora1 logger: Cluster Ready Services waiting on dependencies. Diagnostics in /tmp/crsctl.5782.
May 22 05:30:50 linuxora1 iscsid[3282]: cannot make connection to 192.168.2.196:3260: No route to host
May 22 05:30:50 linuxora1 iscsid[3282]: Connection to Discovery Address 192.168.2.196 failed
The above would be the case as I have emulated the this disk has failed. I have two iscsi devices, the other being 192.168.2.195. These are mounted to the nodes via OCFS2 as /u02 and /u03. Have an OCRfile on each device - plus 2 CRSfiles on /u02 and a CRSfile on /u03
/u03 is the one that is failed. So I should have one OCRfile and 2CRSfiles available.
192.168.2.195 is mounted as /u02 and it's up and running
192.168.2.196 is mounted as /u03 and it currently turned off to emulate a disk failure.
I checked the diagnostics in /tmp/crsctl.5782 as per messages log and it has the following:
[root@linuxora1 bin]# more /tmp/crsctl.5782
OCR initialization failed accessing OCR device: PROC-26: Error while accessing the physical storage
[root@linuxora1 bin]#
So based on this, my problem is that CRS isn't starting up cause OCFS2 can't mount the failed disk? This would explain why the nodes reboot when I emulate the disk failure?

Similar Messages

DPM 2012 R2 fails to back up one of two disks on file server

Good day! I've built a new DPM 2012 R2 server and migrated or re-created jobs from my older 2012 SP1
DPM servers as trying to upgrade them was not working out. The DPM R2 server is running great besides this one issue. The problem is with a file and print server at a remote office. Since it has around 750 GB of user data on it I chose to do manual replica
creation on the data disk, (E:) but created the C:\ drive replica normally, over the WAN. The manual replica creation was successful, and the initial consistency check and recovery point creation worked as expected, however, since then it will not create another
RP or run a CC on the E:\ drive. Recovery points on the C:\ drive are being created properly. When I try to run a CC I get this error: "An unexpected error occurred while the job was running. (ID 104 Details: Access is denied (0x80070005))" I've
tried to drop protection on the two directories of the E:\ drive (retaining data) and re-create, but that didn't help. I've even done a recovery from the initial RP to verify that it was created properly, and that worked as well.

Open command prompt on target file and print server, change directory to
C:\Program Files\Microsoft Data Protection Manager\DPM\Temp> and run below command
C:\Program Files\Microsoft Data Protection Manager\DPM\Temp>find /i "[0x80070005]"
DPMRACur*.errlog >C:\Errors.txt
Paste the output here from C:\Errors.txt

Database control in a two node oracle fail safe cluster

os windows server 2003 r2
db 10.2.0.4
ofs 3.4.1
i've configured db control in the following
D:\oracle\product\10.2.0\db_1\BIN>emca -config dbcontrol db -repos recreate -cluster_name name_of_the_cluster
is this right?
next question
the -cluster flag and the -EM_NODE / -EM_SID_LIST parameters in the emca are only for rac?
with my configuration i can't login to my database over dbcontrol
Listener refused the connection with the following error: ORA-12514, TNS:listener does not currently know of service requested in connect descriptor The Connection descriptor used by the client was: (DESCRIPTION=(ADDRESS_LIST=(ADDRESS=(PROTOCOL=TCP)(HOST=node1.blah)(PORT=1521)))(CONNECT_DATA=(SERVICE_NAME=ABBA)))

How to Configure the 10.2 DB Control in a Failsafe Environment
ID: 452305.1
HOWTO: PUBLISHED 17-OCT-2008
In this Document
 Goal
 Solution
 References
Applies to:
Enterprise Manager Grid Control - Version: 10.2.0.1 to 10.2.0.3
Information in this document applies to any platform.
Goal
The goal of the document is to provide the steps to configure the DB Console in an Oracle Failsafe environment.
Please note that the solution offered in this document is not certified by Oracle. It's only provided as a workaround for customers who wish to use this setup for the Database Control Console in a failsafe environment. Support for the procedure outlined in this document will be provided on a 'best efforts' approach.
As an alternative to this approach you can use Grid Control Agents to monitor a database in a failsafe environment. For more details please see:
Note 396659.1 How To Setup Grid Control Central Agent Manually in Oracle Failsafe Environment
We assume the following:
- the database binaries are installed on each node on the local disk
- you already have a failsafe group with the database, the listener, disk resource, and virtual
ip/hostname
Note: This is specific for Windows environments using Oracle Failsafe. For setting up the DB Console in Active/Passive clusters on UNIX please use the following note instead:
Note 362524.1 Configure DB Control 10.2 for use in Active/Passive High Availability Environments
Solution
1. Drop the DB Control repository and configuration files on the active node:
<ORACLE_HOME>/bin/emca -deconfig dbcontrol db -repos drop
2. Recreate the DB Control using the virtual hostname:
- in a command window do the following:
set oracle_sid=<your_sid>
set oracle_home=<your_oracle_home>
set oracle_hostname=<virtual_hostname>
cd %oracle_home%\bin
emca -config dbcontrol db -repos create
-- make sure that the virtual host name was picked up by emca (it will display a summary screen
before continuing)
3. Verify that the DB Console works fine and can be accessed using the virtual host name in the URL
4. Stop the DB Console
5. Create a directory path on the group disk resource similar to your oracle home path on the local disk.
For example:
If the oracle home on the local disk is C:\oracle\product\10.2.0\db_1 and the shared disk is D:
then create the following path: D:\oracle\product\10.2.0\db_1
Copy C:\oracle\product\10.2.0\db_1\<hostname_sid> and
C:\oracle\product\10.2.0\db_1\oc4j\j2ee\OC4J_DBConsole_<hostname_sid>
to D:\oracle\product\10.2.0\db_1\<hostname_sid> and respectively
D:\oracle\product\10.2.0\db_1\oc4j\j2ee\OC4J_DBConsole_<hostname_sid>
Remember to change the values enclosed in <> with those in your environment.
6. Delete the following directories from the local disk
C:\oracle\product\10.2.0\db_1\<hostname_sid>
C:\oracle\product\10.2.0\db_1\oc4j\j2ee\OC4J_DBConsole_<hostname_sid>
7. Create NTFS junction points from the local disk to the shared disk (these work like symbolic
links on UNIX) . It is your responsibility to create these.
You can find a brief guide as well as some tools that will help you create them at:
http://en.wikipedia.org/wiki/NTFS_junction_point
http://www.microsoft.com/technet ... dDisk/Junction.mspx
Important: Please note that this functionality is not certified nor supported by Oracle. Make sure you understand the
implications. Especially don't delete the junction points from Windows Explorer !!
In the end you should have the junction points on the local disk (with the same names as the
original directories) that would point to the directories you copied on the shared disk.
8. Start the DB Console service and verify that everything is still working fine.
9. Stop the DB Console service.
10. Move the Fail Safe group to the 2nd node.
11. Once the database is started, create the DB Console configuration files on the 2nd node:
<ORACLE_HOME>/bin/emca -config dbcontrol db
12. Stop the DB Console
13. Delete the created configuration files from the local disk:
C:\oracle\product\10.2.0\db_1\<hostname_sid>
C:\oracle\product\10.2.0\db_1\oc4j\j2ee\OC4J_DBConsole_<hostname_sid>
14. Create junction points from the local disk to the directories already located on the shared
disk.
15. Start the DB Console service and verify that is working fine.
16. Add the DB Console service as a resource to the fail safe group and put dependencies on the
database and listener services.
17. Perform a fail over and see if everything works fine.
References
Note 298238.1 - Oracle10g Grid Control with Oracle Fail Safe
Note 362524.1 - Configure DB Control 10.2, 11g for use in Active/Passive High Availability Environments
Note 396659.1 - How To Setup Grid Control Central Agent Manually in Oracle Failsafe Environment
Keywords
EMCA ; DB~CONTROL ; FAILSAFE ; DBCONTROL ; ENTERPRISE~MANAGER ;

RMAN Automatic backup/recovery with oracle fail safe, windows cluster

Hello,
I have question,
1) Whether it is possible to do "RMAN Automatic Backup and recovery" in environment as ?
Environment: -
a. Windows clustering with windows server 2003 Enterprise Edition R2 (Two Node Clustering)
b. Shared disk ( RAID )
c. Oracle 10g standard edition one
d. Oracle fail safe v3.3.3 (for redundancy)
Here we have single oracle instance operating on single database whose files are located on shared disk.
2) If answer to above question is yes please specify if there is some good documentation to it.
Any help regarding this will be greatly appreciated.
Thanks in advance,
Rahul

You just need to make sure that the RMAN scripts are always able to connect to the target database instance whether instance runs on cluster node 1 or cluster node 2
here, If oracle services(resources) will shift from Node1 to Node2 (due to media or any failure) during RMAN Backup then there will be break in connection, will it destroy my backup or will it get started automatically without any harm?
and also I want to know,
Do we need to setup another server which will have RMAN backup script running?

Two Oracle databases on one AIX Server - Note working, Pls Help

Dear Friends,
I have installed Two SAP systems with Two Oracle Databases on one AIX machine, The Installation of both the systems went fine without any errors, and also i have been able to login to both the systems seperately. (that means i cant start both the systems at a time, its one at a time)
But the problem is of the Oracle Listener, i am not able to start both the systems with single Oracle Listener processes(tnslistener). I tried all the possible scenarios as mentioned in note 98252 but nothing seems to be of any use so far.
the thing is i can start either one of the SAP Systems, But my objective is to start both the SAP Systems with one listener processes (LISTENER) and one listener port (1527)
Or
you can say i am also ready to keep two different listeners for TWO seperate Oracle databases, I tried doing so as per one of the sap notes in which you have to keep different names of listener and different numbers for listener port for second ORACLE database. I tried doing exactly mentioned in the note but yet unsuccessful...
Now i am completely lost as what is to be done since its critical i want you to immediately look into it and suggest me the appropriate solution so that i can set up both of my DEVELOPMENT and QUALITY system running...
To troubleshoot this if you require anything from my side then please feel free to ask me since our functional consultants have to soon start working on the SAP systems.
Regards
Ayush

The note you specified talk about 2 database instances not 2 databases... (I was starting to wonder how you deploy 2 databases), you can have multiple database instances on a DB. The note also give you detailed explanation of the configuration of listener.ora
Now, if this is the case and you have installed 2 instances (not databases) on the same host and configured the listener according to the note, then you should be able to call R3trans -d from both SID's.. if the R3trans fails please post the contents of trans.log here.
the thing is i can start either one of the SAP Systems, But my objective is to start both the SAP Systems with one listener processes (LISTENER) and one listener port (1527)
Then, the configuration of your listener.ora file must be incorrect.
Regards
Juan

Installation of oracle fail safe on windows cluster - plz very urgent

hi everybody,
this is the first time i am going to install oracle failsafe on windows clusters.
i am going to install oracle 9i release 2 software.
so i have to install the patches.
plz tell me with oracle cmds how to do the following.
1. after installing oracle 9i rel 2 how to install the patches to it in windows environment?
2. how to create oracle database on one node using shared disk for storage.
3. how to install oracle fail safe on each cluster?
4. how to configure cluster using fail safemanager?
plz provide me soln asap very urgent.
and don't forget to give me the soln with commands aqnd step by step.
plz plz plz
thanks in advance
suresh

For 9i , this is better:
http://download-uk.oracle.com/docs/cd/B10501_01/rac.920/a96600/toc.htm

Win 2003 Cluster + Oracle Fail Safe + Dataguard (physical & Logical)

Hello, 
 
It´s my first post (sorry for my bad english)...I am mounting a high availability solution for test purpose. For the moment i mount the following and runs ok, but i´ve a little problem with the logical database: 
 
Configuration 
ESX Server 2.0 with this machines: 
Windows 2003 Cluster (Enterprise Edition R2, 2 nodes) 
* NODE 1 - Oracle 10gR2 + Patch 9 + Oracle Fail Safe 3.3.4 
* NODE 2 - Oracle 10gR2 + Patch 9 + Oracle Fail Safe 3.3.4 
c:/Windows Software 
e:/Oracle Software/ (pfile -> R:/spfile) 
 
Virtual SAN 
* Datafile, Redos.. are in Virtual SAN. 
R:/ Datafiles & Archivers & dump files & spfile 
S:/ , T:/ ,U:/ -> Redos 
V:/ Undo 
 
Data Guard 
* NODE3 Physical Database 
* NODE4 Logical Database 
 
The Oracle Fail Safe and windows cluster run OK, the switchs... 
The physical database runs OK... (redo aply, switchover, failover, all ok) but the logical receives the redos ok but it has a problem when goes to apply the redo. 
 
The error is the following: 
ORA-12801: error señalizado en el servidor P004 de consultas paralelas 
ORA-06550: linea 1, columna 536: 
PLS-00103: se ha encontrado el simbolo "," cuando se esperaba uno de los siguientes: 
(- + case mod new not null <an identifier>
<a double-quoted delimited-identifier><a bind variable><avg count current exists max min prior sql stddev sum variance execute forall merge time timestamp interval date <a string literal with character set specification><a number> > a single-quoted SQL string> pipe <an alternatively quoted string literal with character set specification> <an alternativel. 
update "SYS"."JOB$" set "LAST_DATE"=TO_DATE('11/09/07','DD/MM/RR'), 
 
This sql statement i saw in dba_logstdby_events and was joined with the error in alert log and dba_logstdby_events. 
 
I´m a bit lost with this error. I don´t understand why the logical database can´t start to apply the redos received from primary database. 
 
The database has two tables with two columns one integer and the other a varchar2(25). She hasn´t rare types of columns. 
 
Thanks a lot for any help, 
Roberto Marotta 

I recreate the logical database OK, no problem, no errors. 
 
The redo aply run ok. I have done logfile switch in primary database and they were applied in logical and standby databases. But... 
 
When I created a tablespace in primary database when i did a switch logfile in primary the changes transfers ok to standby database, but to logical NO!!!, the redo are in they path in logical ok, but when the process tried to apply, reports me the same error. 
 
SQL> select sequence#, first_time, next_time, dict_begin, dict_end, applied from dba_logstdby_log order by 1; 
 
SEQUENCE# FIRST_TI NEXT_TIM DIC DIC APPLIED 
--------- -------- -------- --- --- ------- 
138 14/09/07 14/09/07 NO NO CURRENT 
139 14/09/07 14/09/07 NO NO CURRENT 
 
SQL> select event_time, status, event from dba_logstdby_events order by event_time, timestamp, commit_scn; 
 
14/09/07 
ORA-16222: reintento automatico de la base de datos logica en espera de la ultima accion 
14/09/07 
ORA-16111: extraccion de log y configuracion de aplicacion 
14/09/07 
ORA-06550: linea 1, columna 536: 
PLS-00103: Se ha encontrado el simbolo "," cuando se esperaba uno de los siguientes: 
( - + case mod new not null <an identifier> 
<a double-quoted delimited-identifier> <a bind variable> avg 
count current exists max min prior sql stddev sum variance 
execute forall merge time tiemstamp interval date 
<a string literal with character set specification> 
<a number><a single-quoted SQL string> pipe
<an alternatively-quoted string literal with charactert set specificastion> 
<an alternativel 
update "SYS"."JOB$" set "LAST_NAME" = TO_DATE('14/09/07','DD/MM/RR'), 
 
The alert.log report the same message that the dba_logstdby_events view. 
 
Any idea¿? 
 
I´m a bit frustrated. It´s the third time that recreate the logical database OK and reproduce the same error when i create a tablespace in primary database, and i haven´t got any idea because of that.

Converting Single Instance database into Oracle Fail-Safe on Linux

Hi All,
We have single instance Oracle10g database running on Linux RHEL4. We are looking to convert this database into Oracle Fail-Safe (Active-Passive). Does any one have document for Oracle fail-safe setup?
Regards,
Tushar

Tushar,
you can't use software for windows on Linux.
However, there's number of ways to implement high-availability solution on Linux Oracle.
You can go with Active-Active (RAC) - which means that you'll have two servers attached to the same database at the same time. You can use either, and if one goes down(due to power failure for instance), the second will still be available.
You can go with Active-Passive (Data Guard) - which means that you'll have one of the nodes being active and the second passive, but synchronized with the first one, so in case of failure of the first node you can activate the second one.
You can also go with linux clustering - which is most similar to MCS + fail-safe - you'll have two nodes clustered with linux clustering software (RH cluseter suite for instance) and connected to shared storage. One node is active, and in case of crash database instance will be automatically started on the second node. This is beneficial because you can use your resources more efficiently. In case if you have two databases you can run each of them on dedicated server and then, in case of node failure move to another one. However, the setup of such configuration is quite cumbersome.

Why use oracle fail safe???

hi
i have one question..
why use oracle fail safe???
In the case windows HA System, our systems is operating Oracle database using only Windows Cluster (MSCS) without Oracle Fail Safe!!
Without oracle fail safe, our systems was installed Oracle database.....
if using oracle fail safe, how is merit???.
Thanks...

Hi Kim,
why use oracle fail safe???Refer http://download.oracle.com/docs/html/B12070_01/intro.htm#i1005996
Hope helps
Regrads,
Xaheer

SAP Note 378648 - Offline backup support with Oracle Fail Save

I am trying to implement SAP Note 378648 - Offline backup support with Oracle Fail Save, but am unsure about part of the solution....
Basically, I have to add several parameters into the Environment Variables of the MS Clustered Servers running Windows Server 2003 for the user xidadm...
Fairly straight forward apart from one parameter....
BR_OFS_DOMAIN=<domain>
I am unsure what is required/meant by domain. Can anyone shed some light??

Those are environment variables of the Windows user who is running brbackup; usually this is SAPService<SID>.
You have to set this variables the same way as you set ORACLE_HOME and so on.
(I am not sure about the details; never worked with MS cluster.)
hope this helps

Logs for Oracle Fail Safe

Hi Guys
I need to know where Oracle writes logs for Fail Safe components, Fail Safe Manager and Fail Safe server?
Any idea?
Regards
PG

user10676961 wrote:
First of all, I'm not a native English speaker, so excuse for my msitakes :8}
I have a basic knowlede of Oracle, but this is something I need help with:
I have a Windows cluster with Oracle Fail Safe. There will be multiple databases on this cluster. Oracle and Oracle Fail Safe are installed on disk D: on both nodes (local disk).
When Fail Safe creates a sample database it creates this directory structure on cluster disk O:
OFSDB
OFS1
admin
OFS1
adump
bdump
cdump
pfile
udump
flash_recovery_area
OFS1
ONLINELOG
oradata
OFS1
{code}
All database related files like parameter files are created on the cluster disk, not in Oracle Home, where they are by default.Hi
most of the directories that you refer to are controlled by initialisation parameters in the parameter file for the database. What would probably be sensible for you to do is to use DBCA not to create the database, but to create the scripts to create the database. You can then edit those scripts appropriately.
In particular the directories used in your example are controlled by these parameters
adump                           audit_file_dest
bdump                           background_dump_dest
udump                           user_dump_dest
cdump                           core_dump_dest
flash_recovery_area         db_recovery_file_dest (and db_recovery_file_dest_size)the online log directory tree will be created if you choose to use oracle managed files for the redo logs, alternatively you can manually specify them in the create database statement.
The generated scripts should contain a .bat file which precreates the admin structures.
so to recap, use dbca to create the scripts :- edit the scripts, don't set ORACLE_BASE.
Niall Litchfield
Oracle DBA
http://www.orawin.info/
Edited by: Niall Litchfield on Jan 22, 2009 8:16 AM added code tags for readability

Cluster & Oracle Fail Safe

Hi all! I need to make one instance of Oracle 8 available 24/24 7/7, so I'm searching for a "newbie" guide to configure both Windows Advanced Server and Oracle Fail Safe. Anyone can help me? Any suggests will be appreciated! Thank you very much!

Hi all! I need to make one instance of Oracle 8 available 24/24 7/7, so I'm searching for a "newbie" guide to configure both Windows Advanced Server and Oracle Fail Safe. Anyone can help me? Any suggests will be appreciated! Thank you very much!

Upgrade Oracle Database in Oracle Fail Safe configuration.

Dear All,
I usually upgrade Oracle Database (10g, 10g r2, 11g, 11g r2) in stand alone configuration without problems.
Now, I will must upgrade one database from 11g to 11g r2 in Fail Safe configuration.
Please, is there some link to read before this operation ?
Many thx.

Pl see if MOS Doc 223709.1 (How to upgrade or apply a patch to the RDBMS in an Oracle Fail Safe environment) can help
HTH
Srini

Error in Creating the Oracle Fail Safe Group during installation on MSCS

Hi All,
I am installing ECC 6 dual stack on MSCS.
I have done steps till CI installation as per installation guide.
I am gettig the following error in the step "Creating the Oracle fail safe group" while changing physical hostname into the virtual databse hostname using configtool in secure store parameter jdbc/pool/<SAPSID>Url as per the installation guide.
"Error occurred while working with configuration(scanning)
MSG:Error while connecting to the database"
I think it is not accepting the virtual hostname.
In our scenario CI is running CI node and DB instance is running on DB node.
I have followed the steps as per the installation guide.
Please let me know how to resolve this issue.
Database - Oracle 10g,OS is Windows server 2008 enterprse edition.
Regards,
Dinakar
Edited by: mutnuru dinakar on Jun 1, 2011 1:39 PM
Edited by: mutnuru dinakar on Jun 1, 2011 1:40 PM

Hi,
Thanks for your reply.
1) did you stop the CI before making the modifications as stated in the installation guide?
 Yes
2) is your database currently running on the same physical cluster node as your CI?
 No.Database instance is running on DB node and CI is on CI node.
3) is your database online?
 Yes,it is up and running
Please let me know if there is anyway to resolve this issue.
Regards,
Dinakar

Two disk drives to act as one?

Hello,
I know I might get a LMGTFU for this but what I've looked up so far has been pretty technical and was hoping to get a more simplified explanation. I just got a second hard drive and it's faster and larger than the original that came with it. I know that I can spread out files across two disks but was hoping to find a better solution. What I am looking for is: to bridge the two drives together so that they basically behave as one. I looked at LVM but if I'm right this can do it, however has the caveat of being able to only work with the Linux kernel (Vista takes up one partition). Not knowing anything about it, would it be possible to do some type of RAID configuration to get this functionality? Since I don't know anything about it, I'd appreciate any thoughts you might have. Is this possible (if yes a good place to start), or is this so sort of a pipe dream?

Thanks for the responses. Wasn't aware of the need to reformat. Luckily I got an older third drive that I can swap into my PC and save all the data on. I'd prefer the hardware solution anyways (figure it's easier to set up) and have been looking for a RAID-0 controller and found this:
http://www.newegg.com/Product/Product.a … 6816132001
I only got a PCI slot left, so will this do or am I going to have problems with bottlenecking and such?

Oracle fails if one of two disks fail

Similar Messages

Maybe you are looking for