GoldenGate hands-on in one hour

GoldenGate installation

Take the binary for your platform and database release from Oracle E-Delivery. I have personally taken for my test Oracle GoldenGate V11.1.1.0.0 for Oracle 11g on Linux x86-64.

All testing of this document have been done between two Red Hat Enterprise Linux Server release 5.6 (Tikanga) servers (VMs in fact).

My source and target database are in 11.2.0.2.0 and are both using ASM (which is an additional complexity as we will see).

Then on all servers create a dedicated directory to install the product, I simply chosen to create /goldengate with 1GB size.

[root@server1 /]# lvcreate -n lvol99 -L 1G vg00
  Logical volume "lvol99" created
[root@server1 /]# mkfs -t ext3 -v /dev/vg00/lvol99
mke2fs 1.39 (29-May-2006)
Filesystem label=
OS type: Linux
Block size=4096 (log=2)
Fragment size=4096 (log=2)
131072 inodes, 262144 blocks
13107 blocks (5.00%) reserved for the super user
First data block=0
Maximum filesystem blocks=268435456
8 block groups
32768 blocks per group, 32768 fragments per group
16384 inodes per group
Superblock backups stored on blocks:
        32768, 98304, 163840, 229376
 
Writing inode tables: done
Creating journal (8192 blocks): done
Writing superblocks and filesystem accounting information: done
 
This filesystem will be automatically checked every 34 mounts or
180 days, whichever comes first.  Use tune2fs -c or -i to override.

Edit /etc/fstab and use:

[root@server1 /]# mkdir /goldengate
[root@server1 /]# mount -a
[root@server1 /]# chown orarac:dba /goldengate

Unzip and untar V22230-01.zip file you downloaded.

You may choose to install GoldenGate with a specific account but I have chosen to install it with same account as the one that is owning the database binaries. This avoid environment variables setting, if not your case pay attention to this part as well.

[orarac@server1 goldengate]$ ggsci
 
Oracle GoldenGate Command Interpreter for Oracle
Version 11.1.1.0.0 Build 078
Linux, x64, 64bit (optimized), Oracle 11 on Jul 28 2010 13:13:42
 
Copyright (C) 1995, 2010, Oracle and/or its affiliates. All rights reserved.
 
 
 
GGSCI (server1.domain.com) 1> create subdirs
 
Creating subdirectories under current directory /goldengate
 
Parameter files                /goldengate/dirprm: created
Report files                   /goldengate/dirrpt: created
Checkpoint files               /goldengate/dirchk: created
Process status files           /goldengate/dirpcs: created
SQL script files               /goldengate/dirsql: created
Database definitions files     /goldengate/dirdef: created
Extract data files             /goldengate/dirdat: created
Temporary files                /goldengate/dirtmp: created
Veridata files                 /goldengate/dirver: created
Veridata Lock files            /goldengate/dirver/lock: created
Veridata Out-Of-Sync files     /goldengate/dirver/oos: created
Veridata Out-Of-Sync XML files /goldengate/dirver/oosxml: created
Veridata Parameter files       /goldengate/dirver/params: created
Veridata Report files          /goldengate/dirver/report: created
Veridata Status files          /goldengate/dirver/status: created
Veridata Trace files           /goldengate/dirver/trace: created
Stdout files                   /goldengate/dirout: created
 
 
GGSCI (server1.domain.com) 2> exit
SQL> col VALUE FOR a30
SQL> SET lines 130
SQL> SET pages 100
SQL> SELECT * FROM nls_database_parameters;
 
PARAMETER                      VALUE
------------------------------ ------------------------------
NLS_LANGUAGE                   AMERICAN
NLS_TERRITORY                  AMERICA
NLS_CURRENCY                   $
NLS_ISO_CURRENCY               AMERICA
NLS_NUMERIC_CHARACTERS         .,
NLS_CHARACTERSET               AL32UTF8
NLS_CALENDAR                   GREGORIAN
NLS_DATE_FORMAT                DD-MON-RR
NLS_DATE_LANGUAGE              AMERICAN
NLS_SORT                       BINARY
NLS_TIME_FORMAT                HH.MI.SSXFF AM
NLS_TIMESTAMP_FORMAT           DD-MON-RR HH.MI.SSXFF AM
NLS_TIME_TZ_FORMAT             HH.MI.SSXFF AM TZR
NLS_TIMESTAMP_TZ_FORMAT        DD-MON-RR HH.MI.SSXFF AM TZR
NLS_DUAL_CURRENCY              $
NLS_COMP                       BINARY
NLS_LENGTH_SEMANTICS           BYTE
NLS_NCHAR_CONV_EXCP            FALSE
NLS_NCHAR_CHARACTERSET         AL16UTF16
NLS_RDBMS_VERSION              11.2.0.2.0
 
20 ROWS selected.

Then simply issue (or better add it in your .profile, .bash_profile…):

[orarac@server1 goldengate]$ export NLS_LANG=AMERICAN_AMERICA.AL32UTF8

Then on source database we create a test table with:

SQL> CREATE TABLE test1 AS SELECT * FROM dba_tables;
 
TABLE created.

Remark:
You must create the structure of the table on target database to be able to use the initial loading. Strangely the initial loading will load for you the data but will NOT create the table.

At Oracle database level you must activate supplemental logging on source database with:

SQL> ALTER DATABASE ADD supplemental LOG data;
 
DATABASE altered.
 
SQL> ALTER SYSTEM switch logfile;
 
SYSTEM altered.
 
SQL> SELECT supplemental_log_data_min FROM v$database;
 
SUPPLEME
--------
YES

And have your source database in archivelog mode !

the target database can even be in noarchivelog mode.

In your Oracle database home directory create an alias to access your ASM instance like the following:

+ASM.WORLD =
  (DESCRIPTION =
    (ADDRESS_LIST =
      (ADDRESS = (PROTOCOL = TCP)(Host = server1.domain.com)(Port = 1531))
    )
    (CONNECT_DATA =
      (SID = +ASM)
      (SERVER = DEDICATED)
    )
  )

Manager processes

As stated in official documentation manager process must run on source and target server and helps to (you start one manager process than can manage one or more extract or replicat processes):

  • Start Oracle GoldenGate processes
  • Start dynamic processes
  • Perform trail management
  • Create event, error, and threshold reports

Create the manager parameter file with (do same action on source and target servers):

GGSCI (server1.domain.com) 1> edit param mgr

Then enter the following parameters (and anyone else you want):

port 7809
laginfoseconds 60
purgeoldextracts /goldengate/dirdat/aa*
autorestart er capture
autorestart er stage
autostart er capture
autostart er stage

Parameters:

  • port: port of the manager process.
  • laginfoseconds: how often you want lag information to be reported in error log file (laginfominutes, laginfohours). I have not been able to find any log file but apparently this parameter is need to have relevant information in Lag column when issuing info all command.
  • purgeoldextracts: purge trail files when GoldenGate has finished processing them.
  • autostart and autorestart are self explaining. Be careful when using them as it will also restart the loading extract/replicat if you use * for group name…

Start the manager with:

GGSCI (server1.domain.com) 2> info mgr
 
Manager is DOWN!
 
 
GGSCI (server1.domain.com) 3> start mgr
 
Manager started.
 
GGSCI (server1.domain.com) 4> info mgr
 
Manager is running (IP port server1.domain.com.7809).
 
 
GGSCI (server1.domain.com) 5> view report mgr
 
 
***********************************************************************
                 Oracle GoldenGate Manager for Oracle
                     Version 11.1.1.0.0 Build 078
   Linux, x64, 64bit (optimized), Oracle 11 on Jul 28 2010 13:12:40
 
Copyright (C) 1995, 2010, Oracle and/or its affiliates. All rights reserved.
 
 
                    Starting at 2011-05-09 17:26:03
***********************************************************************
 
Operating System Version:
Linux
Version #1 SMP Tue Jan 18 20:11:49 EST 2011, Release 2.6.32-100.26.2.el5
Node: server1.domain.com
Machine: x86_64
                         soft limit   hard limit
Address Space Size   :    unlimited    unlimited
Heap Size            :    unlimited    unlimited
File Size            :    unlimited    unlimited
CPU Time             :    unlimited    unlimited
 
Process id: 4796
 
Parameters...
 
port 7809
laginfoseconds 60
purgeoldextracts /goldengate/dirdat/aa*
 
 
***********************************************************************
**                     Run Time Messages                             **
***********************************************************************
 
 
2011-05-09 17:26:03  INFO    OGG-00983  Manager started (port 7809).

Create gss account on both databases:

SQL> CREATE USER ggs IDENTIFIED BY ggs;
 
USER created.

Then on source database (extract):

SQL> GRANT CREATE SESSION, ALTER SESSION, RESOURCE, CONNECT, SELECT ANY dictionary TO ggs;
 
GRANT succeeded.
 
SQL> GRANT flashback ANY TABLE, SELECT ANY TABLE TO ggs;
 
GRANT succeeded.
 
SQL> GRANT EXECUTE ON dbms_flashback TO ggs;
 
GRANT succeeded.

On target (Replicat) database:

SQL> GRANT CREATE SESSION, ALTER SESSION, RESOURCE, CONNECT, SELECT ANY dictionary TO ggs;
 
GRANT succeeded.
 
SQL> GRANT SELECT ANY TABLE, INSERT ANY TABLE, UPDATE ANY TABLE, DELETE ANY TABLE TO ggs;
 
GRANT succeeded.
 
SQL> GRANT CREATE TABLE, ALTER ANY TABLE, LOCK ANY TABLE TO ggs;
 
GRANT succeeded.

Add supplemental log to tables you want to replicate (same as Streams):

[orarac@server1 goldengate]$ ggsci
 
Oracle GoldenGate Command Interpreter for Oracle
Version 11.1.1.0.0 Build 078
HP/UX, IA64, 64bit (optimized), Oracle 10 on Jul 28 2010 14:36:33
 
Copyright (C) 1995, 2010, Oracle and/or its affiliates. All rights reserved.
 
 
 
GGSCI (server1.domain.com) 1> dblogin, userid ggs, password ggs
Successfully logged into database.
 
GGSCI (server1.domain.com) 2> add trandata yjaquier.test1
 
2011-04-21 14:16:01  WARNING OGG-00869  No unique key is defined for table TEST1. All viable columns will be used to represent the key, but may not guarantee uniqueness.  KEYCOLS may be used to define the key.
 
2011-04-21 14:16:01  WARNING OGG-01387  Table YJAQUIER.TEST1 has no valid key columns, added unconditional supplemental log group for all table columns.
 
2011-04-21 14:16:01  WARNING OGG-00706  Failed to add supplemental log group on table YJAQUIER.TEST1 due to ORA-01031: insufficient privileges, SQL ALTER TABLE "YJAQUIER"."TEST1" ADD SUPPLEMENTAL LOG DATA (ALL) COLUMNS  /* GOLDENGATE_DDL_REPLICATION */.

Despite GoldenGate documentation one grant is missing (that’s probably why Oracle grant DBA to this account):

SQL> GRANT ALTER ANY TABLE TO ggs;
 
GRANT succeeded.

Relaunch last command:

GGSCI (server1.domain.com) 3> add trandata yjaquier.test1
 
2011-04-21 14:17:29  WARNING OGG-00869  No unique key is defined for table TEST1. All viable columns will be used to represent the key, but may not guarantee uniqueness.  KEYCOLS may be used to define the key.
 
2011-04-21 14:17:29  WARNING OGG-01387  Table YJAQUIER.TEST1 has no valid key columns, added unconditional supplemental log group for all table columns.
 
Logging of supplemental redo data enabled for table YJAQUIER.TEST1.

Remark:
As you see just above it is strongly suggested to have a primary key on the replicated table, it speed up the apply process (same as Data Guard logical standby).

You would do the same with an ALTER TABLE statement and you can verify it has been done at Oracle level:

SQL> SET lines 200
SQL> SELECT * FROM dba_log_groups ORDER BY owner, table_name,log_group_type WHERE owner='YJAQUIER';
 
OWNER                          LOG_GROUP_NAME                 TABLE_NAME                     LOG_GROUP_TYPE               ALWAYS      GENERATED
------------------------------ ------------------------------ ------------------------------ ---------------------------- ----------- --------------
YJAQUIER                       SYS_C0011067                   TEST1                          ALL COLUMN LOGGING           ALWAYS      GENERATED NAME

Initial loading

Definition file

First we will create table definition file. This readable file contains the format of data that will be synchronized, please refer to official documentation for added value of working with definition file:

GGSCI (server1.domain.com) 4> edit param defgen

Create below file with vi:

defsfile /goldengate/dirdef/source.def, purge
userid ggs, password ggs
table yjaquier.test1;

Execute it with:

[orarac@server1 goldengate]$ defgen paramfile ./dirprm/defgen.prm
 
***********************************************************************
        Oracle GoldenGate Table Definition Generator for Oracle
                     Version 11.1.1.0.0 Build 078
   Linux, x64, 64bit (optimized), Oracle 11 on Jul 28 2010 13:15:49
 
Copyright (C) 1995, 2010, Oracle and/or its affiliates. All rights reserved.
 
 
                    Starting at 2011-05-09 17:52:41
***********************************************************************
 
Operating System Version:
Linux
Version #1 SMP Tue Jan 18 20:11:49 EST 2011, Release 2.6.32-100.26.2.el5
Node: server1.domain.com
Machine: x86_64
                         soft limit   hard limit
Address Space Size   :    unlimited    unlimited
Heap Size            :    unlimited    unlimited
File Size            :    unlimited    unlimited
CPU Time             :    unlimited    unlimited
 
Process id: 5387
 
***********************************************************************
**            Running with the following parameters                  **
***********************************************************************
defsfile /goldengate/dirdef/source.def, purge
userid ggs, password ***
table yjaquier.test1;
Retrieving definition for YJAQUIER.TEST1
 
2011-05-09 17:52:41  WARNING OGG-00869  No unique key is defined for table TEST1. All viable columns will be used to represent the key, but may not guarantee uniqueness.  KEYCOLS may be used to define the key.
 
 
Definitions generated for 1 tables in /goldengate/dirdef/source.def

Copy this definition file to target server:

[orarac@server1 dirdef]$ pwd
/goldengate/dirdef
[orarac@server1 dirdef]$ scp source.def orarac@server2:/goldengate/dirdef
orarac@server2's password:
source.def

Replication process on source server

Create the replication process (8 characters maximum !!, ERROR: Invalid group name (must be at most 8 characters).), extract part. This process will read figures on source server and transfer them to remote server:

GGSCI (server1.domain.com) 1> edit param srv1srv2
 
 
GGSCI (server1.domain.com) 2> shell cat /goldengate/dirprm/srv1srv2.prm
 
extract srv1srv2
userid ggs, password ggs
rmthost server2.domain.com, mgrport 7809
rmttask replicat, group srv1srv2
table yjaquier.test1;

Then create the command to launch it:

[orarac@server1 goldengate]$ pwd
/goldengate
[orarac@server1 goldengate]$ mkdir diroby
[orarac@server1 goldengate]$ vi ./diroby/srv1srv2.oby
[orarac@server1 goldengate]$ cat ./diroby/srv1srv2.oby
add extract srv1srv2, sourceistable

Remark:
Sourceistable is the keyword to create an initial loading extract and not an online synchronization process.

Replication process on target server

Do the same on target server. This process will read figures and insert them in target tables:

GGSCI (server2.domain.com) 1> edit param srv1srv2
 
GGSCI (server2.domain.com) 2> shell cat "/goldengate/dirprm/srv1srv2.prm"
 
replicat srv1srv2
userid ggs, password ggs
sourcedefs /goldengate/dirdef/source.def
dboptions deferrefconst
map yjaquier.*, target yjaquier.*;

Remark:
Dboptions deferrefconst simply delays integrity constraints validation until commit, particularly useful when working with Oracle (valid if Oracle > 9.2.0.7).

Create as well the start command script:

[orarac@server2 goldengate]$ pwd
/goldengate
[orarac@server2 goldengate]$ mkdir ./diroby
[orarac@server2 goldengate]$ vi ./diroby/srv1srv2.oby
[orarac@server2 goldengate]$ cat ./diroby/srv1srv2.oby
add replicat srv1srv2, specialrun

Remark:
Specialrun is the counter part of sourceistable and is used only when creating the initial loading replicat.

Start first the replication on target server with:

GGSCI (server2.domain.com) 1> obey ./diroby/srv1srv2.oby
 
GGSCI (server2.domain.com) 2> add replicat srv1srv2, specialrun
 
REPLICAT added.
 
 
GGSCI (server2.domain.com) 3> start replicat srv1srv2
 
Sending START request to MANAGER ...
REPLICAT SRV1SRV2 starting
 
 
GGSCI (server2.domain.com) 4> info srv1srv2
 
REPLICAT   SRV1SRV2  Initialized   2011-05-09 18:03   Status STOPPED
Checkpoint Lag       00:00:00 (updated 00:00:52 ago)
Log Read Checkpoint  Not Available
Task                 SPECIALRUN

Then same on source server:

GGSCI (server1.domain.com) 1> obey ./diroby/srv1srv2.oby
 
GGSCI (server1.domain.com) 2> add extract srv1srv2, sourceistable
 
EXTRACT added.
 
 
GGSCI (server1.domain.com) 3> start extract srv1srv2
 
Sending START request to MANAGER ...
EXTRACT SRV1SRV2 starting
 
 
GGSCI (server1.domain.com) 4> info srv1srv2
 
EXTRACT    SRV1SRV2  Initialized   2011-05-09 18:05   Status STOPPED
Checkpoint Lag       Not Available
Log Read Checkpoint  Not Available
                     First Record         Record 0
Task                 SOURCEISTABLE

Then if you check on target database your target TEST1 table has been populated !!

Data synchronization

The process is the same as Oracle Streams (CSA):

  • Capture
  • Stage (or Pump)
  • Apply

Remark:
We will use those names to create the extract and replicat processes.

Capture and stage processes

On source server we modify parameters of capture and stage processes:

GGSCI (server1.domain.com) 1> edit param capture
 
 
 
GGSCI (server1.domain.com) 2> shell cat /goldengate/dirprm/capture.prm
 
extract capture
userid ggs, password ggs
tranlogoptions asmuser sys@+ASM, asmpassword password
exttrail /goldengate/dirdat/aa
table yjaquier.test1;
 
 
GGSCI (server1.domain.com) 3> edit param stage
 
 
 
GGSCI (server1.domain.com) 4> shell cat /goldengate/dirprm/stage.prm
 
extract stage
rmthost server2.domain.com, mgrport 7809
rmttrail /goldengate/dirdat/aa
passthru
table yjaquier.test1;

Remarks:
Tranlogoptions must be used if your archived log files are stored with ASM. Passthru speed up replication process when no data manipulation are need on objects.

Now we need to create and start the two processes we have just modified parameters. This will be done using an obey script:

GGSCI (server1.domain.com) 1> shell vi ./diroby/extract.oby
 
 
 
GGSCI (server1.domain.com) 2> shell cat ./diroby/extract.oby
 
kill extract capture
delete extract capture
add extract capture, tranlog, begin now
add exttrail /goldengate/dirdat/aa, megabytes 1, extract capture
start extract capture
kill extract stage
delete extract stage
add extract stage, exttrailsource /goldengate/dirdat/aa
add rmttrail /goldengate/dirdat/aa, megabytes 1, extract stage
start stage

Remark:
The script creates and starts capture and stage processes. We simply limit trail files to 1MB.

Apply process

On target system we will modify parameters of apply process:

GGSCI (server2.domain.com) 1> edit param apply
 
 
 
GGSCI (server2.domain.com) 2> shell cat /goldengate/dirprm/apply.prm
 
replicat apply
userid ggs, password ggs
discardfile /goldengate/dirout/apply.dsc, append
sourcedefs /goldengate/dirdef/source.def
insertmissingupdates
dboptions deferrefconst
map yjaquier.test1, target yjaquier.test1, colmap (usedefaults);

Remark:
Insertmissingupdates that is not default value will not trigger an error when target records do not exist. It is expected behavior in my opinion so strange that it is not default value… The map keyword is a pretty complex one and allow you to perform any kind of transformations you may think.

Then same as before we will create an oby script to define and start the apply process:

GGSCI (server2.domain.com) 1> shell vi ./diroby/apply.oby
 
 
 
GGSCI (server2.domain.com) 2> shell cat ./diroby/apply.oby
 
kill extract apply
delete replicat apply
add replicat apply, exttrail /goldengate/dirdat/aa, begin now, nodbcheckpoint
start replicat apply

Remark:
Nodbcheckpoint does not log information into checkpoint table. you can specify one using checkpointtable . The script chkpt_ora_create.sql that is under GoldenGate home directory create one called ggs_checkpoint (ideal owner is obviously ggs). Note that it can be set at global level using edit params /goldengate/GLOBALS. Then at Oracle level you can see it with:

SQL> SET lines 200
SQL> SELECT * FROM ggs.ggs_checkpoint;
 
GROUP_NA  GROUP_KEY      SEQNO        RBA AUDIT_TS                      CREATE_TS          LAST_UPDATE_TS
-------- ---------- ---------- ---------- ----------------------------- ------------------ ------------------
CURRENT_DIR
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
APPLY    2253737335          5      28024 2011-05-11 14:12:59.000000    11-MAY-11          11-MAY-11
/goldengate

Then we start on source server the capture and stage processes:

GGSCI (server1.domain.com) 1> obey /goldengate/diroby/extract.oby
 
GGSCI (server1.domain.com) 2> kill extract capture
 
ERROR: Invalid command.
 
GGSCI (server1.domain.com) 3> delete extract capture
 
ERROR: EXTRACT CAPTURE does not exist.
 
 
GGSCI (server1.domain.com) 4> add extract capture, tranlog, begin now
 
EXTRACT added.
 
 
GGSCI (server1.domain.com) 5> add exttrail /goldengate/dirdat/aa, megabytes 1, extract capture
 
EXTTRAIL added.
 
 
GGSCI (server1.domain.com) 6> start extract capture
 
 
Sending START request to MANAGER ...
EXTRACT CAPTURE starting
 
 
GGSCI (server1.domain.com) 7> kill extract stage
 
ERROR: Invalid command.
 
GGSCI (server1.domain.com) 8> delete extract stage
 
ERROR: EXTRACT STAGE does not exist.
 
 
GGSCI (server1.domain.com) 9> add extract stage, exttrailsource /goldengate/dirdat/aa
 
EXTRACT added.
 
 
GGSCI (server1.domain.com) 10> add rmttrail /goldengate/dirdat/aa, megabytes 1, extract stage
 
RMTTRAIL added.
 
 
GGSCI (server1.domain.com) 11> start stage
 
 
Sending START request to MANAGER ...
EXTRACT STAGE starting
 
 
GGSCI (server1.domain.com) 12> info all
 
Program     Status      Group       Lag           Time Since Chkpt
 
MANAGER     RUNNING
EXTRACT     RUNNING     CAPTURE     00:00:00      00:00:08
EXTRACT     RUNNING     STAGE       00:00:00      00:00:06
 
 
GGSCI (server1.domain.com) 13> info *
 
EXTRACT    CAPTURE   Last Started 2011-05-10 17:14   Status RUNNING
Checkpoint Lag       00:00:00 (updated 00:00:05 ago)
Log Read Checkpoint  Oracle Redo Logs
                     2011-05-10 18:00:27  Seqno 94, RBA 10998784
 
EXTRACT    STAGE     Last Started 2011-05-10 16:29   Status RUNNING
Checkpoint Lag       00:00:00 (updated 00:00:02 ago)
Log Read Checkpoint  File /goldengate/dirdat/aa000000
                     2011-05-10 16:57:23.000000  RBA 1547
 
 
GGSCI (server1.domain.com) 14> info capture, detail
 
EXTRACT    CAPTURE   Last Started 2011-05-17 14:30   Status RUNNING
Checkpoint Lag       00:00:00 (updated 00:00:01 ago)
Log Read Checkpoint  Oracle Redo Logs
                     2011-05-17 15:00:21  Seqno 99, RBA 12639232
 
  Target Extract Trails:
 
  Remote Trail Name                                Seqno        RBA     Max MB
 
  /goldengate/dirdat/aa                                0          0          1
 
  Extract Source                          Begin             End
 
  +DATA/rac/onlinelog/group_3.262.744732371  2011-05-16 16:27  2011-05-17 15:00
  +DATA/rac/onlinelog/group_2.264.744732365  2011-05-11 17:43  2011-05-16 16:27
  +DATA/rac/onlinelog/group_1.266.744732363  2011-05-11 14:36  2011-05-11 17:43
  Not Available                           * Initialized *   2011-05-11 14:36
 
 
Current directory    /goldengate
 
Report file          /goldengate/dirrpt/CAPTURE.rpt
Parameter file       /goldengate/dirprm/capture.prm
Checkpoint file      /goldengate/dirchk/CAPTURE.cpe
Process file         /goldengate/dirpcs/CAPTURE.pce
Stdout file          /goldengate/dirout/CAPTURE.out
Error log            /goldengate/ggserr.log
 
 
GGSCI (server1.domain.com) 15> lag capture
 
Sending GETLAG request to EXTRACT CAPTURE ...
Last record lag: 1 seconds.
At EOF, no more records to process.

Then we start the apply process on target server:

GGSCI (server2.domain.com) 1> start /goldengate/diroby/apply.oby
ERROR: Invalid command.
 
GGSCI (server2.domain.com) 2> obey /goldengate/diroby/apply.oby
 
GGSCI (server2.domain.com) 3> kill extract apply
 
ERROR: Invalid command.
 
GGSCI (server2.domain.com) 4> delete replicat apply
 
ERROR: REPLICAT APPLY does not exist.
 
 
GGSCI (server2.domain.com) 5> add replicat apply, exttrail /goldengate/dirdat/aa, begin now, nodbcheckpoint
 
REPLICAT added.
 
 
GGSCI (server2.domain.com) 6> start replicat apply
 
 
Sending START request to MANAGER ...
REPLICAT APPLY starting
 
 
GGSCI (server2.domain.com) 7> info all
 
Program     Status      Group       Lag           Time Since Chkpt
 
MANAGER     RUNNING
REPLICAT    RUNNING     APPLY       00:00:00      00:00:04
 
 
GGSCI (server2.domain.com) 8> info *
 
REPLICAT   APPLY     Last Started 2011-05-10 15:13   Status RUNNING
Checkpoint Lag       00:00:00 (updated 00:00:09 ago)
Log Read Checkpoint  File /goldengate/dirdat/aa000000
                     2011-05-10 16:57:18.925791  RBA 1591
 
 
GGSCI (server2.domain.com) 9> info apply, detail
 
REPLICAT   APPLY     Last Started 2011-05-17 14:31   Status RUNNING
Checkpoint Lag       00:00:00 (updated 00:00:05 ago)
Log Read Checkpoint  File /goldengate/dirdat/aa000008
                     2011-05-17 14:30:34.409181  RBA 1493
 
  Extract Source                          Begin             End
 
  /goldengate/dirdat/aa000008             2011-05-16 14:49  2011-05-17 14:30
  /goldengate/dirdat/aa000007             2011-05-11 14:36  2011-05-16 14:49
  /goldengate/dirdat/aa000006             2011-05-11 14:12  2011-05-11 14:36
  /goldengate/dirdat/aa000000             * Initialized *   2011-05-11 14:12
 
 
Current directory    /goldengate
 
Report file          /goldengate/dirrpt/APPLY.rpt
Parameter file       /goldengate/dirprm/apply.prm
Checkpoint file      /goldengate/dirchk/APPLY.cpr
Checkpoint table     GGS.GGS_CHECKPOINT
Process file         /goldengate/dirpcs/APPLY.pcr
Stdout file          /goldengate/dirout/APPLY.out
Error log            /goldengate/ggserr.log

Testing

SQL> SELECT TABLE_NAME FROM test1 WHERE owner='YJAQUIER';
 
TABLE_NAME
------------------------------
T1
T3
 
SQL> INSERT INTO test1(OWNER,TABLE_NAME) VALUES('YJAQUIER','TEST1');
 
1 ROW created.
 
SQL> COMMIT;
 
COMMIT complete.
 
SQL> SELECT TABLE_NAME FROM test1 WHERE owner='YJAQUIER';
 
TABLE_NAME
------------------------------
T1
T3
TEST1

And magically on target server:

SQL> SELECT TABLE_NAME FROM test1 WHERE owner='YJAQUIER';
 
TABLE_NAME
------------------------------
T1
T3
TEST1

Error correction

Missing archived log file

Error is something like:

2011-05-10 15:13:23  ERROR   OGG-00446  Could not find archived log for sequence 92 thread 1 under default destinations SQL <SELECT  name    FROM v$archived_log   WHERE sequence# = :ora_seq_no AND         th
read# = :ora_thread AND         resetlogs_id = :ora_resetlog_id AND         archived = 'YES' AND         deleted = 'NO>, error retrieving redo file name for sequence 92, archived = 1, use_alternate = 0Not ab
le to establish initial position for begin time 2011-05-10 14:38:39.

To solve it (and accept being obliged to re-conciliate figures) you can issue:

GGSCI (server1.domain.com) 1> alter extract capture, extseqno 93
EXTRACT altered.
 
 
GGSCI (server1.domain.com) 2> start capture
 
Sending START request to MANAGER ...
EXTRACT CAPTURE starting

Archived log in ASM

Error looks like:

2011-05-10 16:57:56  ERROR   OGG-00446  Could not find archived log for sequence 93 thread 1 under default destinations SQL <SELECT  name    FROM v$archived_log   WHERE sequence# = :ora_seq_no AND         th
read# = :ora_thread AND         resetlogs_id = :ora_resetlog_id AND         archived = 'YES' AND         deleted = 'NO>, error retrieving redo file name for sequence 93, archived = 1, use_alternate = 0Not ab
le to establish initial position for sequence 93, rba 0.

While you see it at Oracle level:

SQL> SET lines 200
SQL>  col name FOR a75
SQL> SELECT SEQUENCE#, name FROM v$archived_log;
 
 SEQUENCE# NAME
---------- ---------------------------------------------------------------------------
        93 +DATA/rac/archivelog/2011_05_10/thread_1_seq_93.270.750784485

Use the tranlogoptions asmuser sys@+ASM, asmpassword paswowrd option for the capture process.

Figures not present in target table

Error is something like:

2011-05-10 18:07:30  ERROR   OGG-01296  Error mapping from YJAQUIER.TEST1 to YJAQUIER.TEST1.
 
***********************************************************************
*                   ** Run Time Statistics **                         *
***********************************************************************
 
Last record for the last committed transaction is the following:
___________________________________________________________________
Trail name :  /goldengate/dirdat/aa000000
Hdr-Ind    :     E  (x45)     Partition  :     .  (x04)
UndoFlag   :     .  (x00)     BeforeAfter:     B  (x42)
RecLength  :   470 (x01d6)    IO Time    : 2011-05-10 18:05:56.765745
IOType     :     3  (x03)     OrigNode   :   255  (xff)
TransInd   :     .  (x03)     FormatType :     R  (x52)
SyskeyLen  :     0  (x00)     Incomplete :     .  (x00)
AuditRBA   :         94       AuditPos   : 11454992
Continued  :     N  (x00)     RecCount   :     1  (x01)
 
2011-05-10 18:05:56.765745 Delete             Len   470 RBA 1591
Name: YJAQUIER.TEST1
___________________________________________________________________
 
Reading /goldengate/dirdat/aa000000, current RBA 1591, 0 records
 
Report at 2011-05-10 18:07:30 (activity since 2011-05-10 18:07:30)
 
From Table YJAQUIER.TEST1 to YJAQUIER.TEST1:
       #                   inserts:         0
       #                   updates:         0
       #                   deletes:         0
       #                  discards:         1

You can confirm the error by looking at the discard file you configured in apply process:

[orarac@server2 dirout]$ pwd 
/goldengate/dirout
[orarac@server2 dirout]$ head -20 apply.dsc
Oracle GoldenGate Delivery for Oracle process started, group APPLY discard file opened: 2011-05-10 18:07:22
 
Current time: 2011-05-10 18:07:30
Discarded record from action ABEND on error 1403
 
 
Aborting transaction on /goldengate/dirdat/aa beginning at seqno 0 rba 1591
                         error at seqno 0 rba 1591
Problem replicating YJAQUIER.TEST1 to YJAQUIER.TEST1
Record not found
Mapping problem with delete record (target format)...
*
OWNER = YJAQUIER
TABLE_NAME = TEST1
TABLESPACE_NAME = NULL
CLUSTER_NAME = NULL
IOT_NAME = NULL
STATUS = NULL
PCT_FREE = NULL
PCT_USED = NULL

One option to solve it could be to add handlecollisions in your apply parameters file (it may be your default option if you don’t want your apply process to be down each time users are doing *bad* updates):

GGSCI (server2.domain.com) 41> edit param apply
 
 
GGSCI (server2.domain.com) 45> shell cat ./dirprm/apply.prm
 
replicat apply
userid ggs, password ggs
discardfile /goldengate/dirout/apply.dsc, purge
sourcedefs /goldengate/dirdef/source.def
insertmissingupdates
dboptions deferrefconst
handlecollisions
map yjaquier.test1, target yjaquier.test1, colmap (usedefaults);

References

About Post Author

This entry was posted in Oracle and tagged . Bookmark the permalink.

16 thoughts on “GoldenGate hands-on in one hour

  1. Hi Yannick,

    did you find a situation, where data missing on the target server, as we have recently implemented golden gate, but we have got this issue couple of times, i was late on your site.

    Please help me out if you have anything in your mind regarding this.

    Jatti

    • Hi Jatti,

      We finally did not deploy GoldenGate under complex environment so I have poor return on real life example. We are going to have a massive deployment early next year I might come back with few ideas at that time. But I hope you will have solved your issue before that… Also in meanwhile GoldenGate has evolved so maybe now everything is more under control…

      Thanks, Yannick.

  2. Thanks,
    Yes, TABLE keyword is fine. We have 70 tables to capture and apply change in golden gate from a schema having 1300 odd tables.
    My question is how to use tokens keyword once for 70 tables. Currently, I can make it work by using TOKENS keyword for each table separately making extract file very big.
    Is there a way I can use TOKENS keyword once. I can use TABLEEXCLUSION keyword and table hq.* but that is maintenance nightmare. Will have to add TABLEEXLUSION for every new table.
    Thanks, gusi

  3. Hi,
    Good article.
    How can we use multiple tables for the same schema (not all tables of the schema but subset of tables) in the extract file
    I was able to find that I can use this in the extract file:
    TABLE hq.acct;
    TABLE hq.sales;

    My question is how to use tokens after this since we are populating hostname etc. Do I need to use Tokens command for each table defined in the capture parameter file?
    thanks, gusi

  4. HI Yannick,
    thanks for your article it is very useful, i have a question please can we copy the trial file generated from source GoldenGate db to target GoldenGate db manually, so goldengate can apply them as it is moved automatically? …
    i’m talking in case the network connections goes down and we need GoldenGate to keep working …
    please advise

    Thank You

    • Hi Mazen,

      Thanks for stopping by !

      I would say it should work but I have not tested it and have scratched my test environment.

      If you refer to OGG Extract TCP/IP Performance troubleshooting (Doc ID 1428353.1) MOS note then manually copying trail files is what Oracle suggest to troubleshoot network performance issues.

      Thanks, Yannick.

  5. Chandra Cheedella says:

    Dear Yannick,
    Thank you so much for writing this wonderful article. Its impressive with all the needed details in it and I feel very comfortable now with GG setup & general operations. Job well done.
    With regards.

  6. Hi,
    This is the issue have, we have what I will call the database “A” ( it’s an Oracle DB ), that we developers perform our (CRUD) operation. In addition we have the Oracle Goldengate product that our DBA’s setup. With this product they configured it to perform the “Change Data Capture” on database “A” and store the changes in database “B”. This was all done but configuration in Goldengate. The issue now has come up that there is no process to purge database “B”, as the volume is just growing with “Change Data Capture” operations done by the CRUD operations in database “A”. Our so called DBA is now saying that we developers have to write the code to purge database “B”, which contains about 650 tables. (sorry it’s not a well-designed database), but I am trying to push back and say to upper management that this operation should be configured in Goldengate. So my question is, Dose Goldengate have a configuration to purge the database “B”.
    Thanks for your help.

    • Hi,

      Not sure to catch your point. If data are not purged in database B but are in database A it means you have a particular configuration using something similar to IGNOREDELETES. So on database B you must implement an “applicative” purging of data based on business rules (after 6 months, …) and for me it has nothing to see with GoldenGGate…

      Thanks, Yannick.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes:

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>