Many redo logs switches the line seconds on an irregular basis

Hello

We are 11 GR 2 RAC. We migrated a DB 10 g that used to pass on every 20-25 minutes all the

While doing about 160 to 200 GB per day; I noticed this newly migrated DB now show

pics of quick saves switches from time to time, for example yesterday (excerpt):

SQL > select thread #, FIRST_TIME from v$ log_history by 2;

...

1 20140524 10:19.36
1 20140524 11:24.34
2 20140524 11:24.35
2 20140524 12:00.02
1 20140524 12:00.03
1 20140524 12:00.05
2 20140524 12:00.08
1 20140524 12:00.14
1 20140524 13:06.19
2 20140524 13:09.44
2 20140524 14:15.46
1 20140524 15:00.08
2 20140524 15:00.50
2 20140524 15:53.56
2 20140524 16:45.37
1 20140524 16:45.39
2 20140524 17:43.26

Ignore lines because of lunch at this time that I start a backup of archive logs, but to explain

switches 11:24.34 et.35 for example, or 15:00.08 et.50, then at 16:45, where

Do you think I should study? I have no reason why in these moments my DB suddenly

give 3 GB of Oder, before you wait another hour to do this (e.g. between 16:45 and 17:43)...

Everything about the fact that we are on CARS now?

Thank you very much.

Kind regards

SEB

Nothing to worry about.

Each node in a RAC has it of own redo logs and performs his own logswitches.

If you look for each thread (1 and 2 that correspond to the node) itself, everything is ok.

Tags: Database

Similar Questions

  • Redo log switch always calls the control point?

    Hello

    Is a log switch redo always calls process checkpoint?

    I have a database of test on which nothing is not running, but when I switch the redo log file, so it takes time to make the State of redo log active to INACTIVE.

    But when I change check system; the State immediately changes from ACTIVE to INACTIVE.

    Gives me a feeling that the redo log switch, it is not mandatory for the checkpoint occurs.

    DB version is 11.2.0.1

    Comment of nicely.

    A 11.2.0.4 alert after log is 3 'alter system switch logfile' in a line:

    ALTER SYSTEM SET log_checkpoints_to_alert = TRUE SCOPE = BOTH;

    Wed Apr 16 22:17:09 2014

    Beginning log switch checkpoint up to RBA [0xd8.2.10], SNA: 12670277595657 *.

    Thread 1 Advanced to save the 216 (switch LGWR) sequence

    Currently Journal # 3 seq # 216 mem # 0: /u01/app/oracle/oradata/TEST/onlinelog/o1_mf_3_938s43lb_.log

    Currently Journal # 3 seq # 216 mem # 1: /u01/app/oracle/fast_recovery_area/TEST/onlinelog/o1_mf_3_938s49xz_.log

    Wed Apr 16 22:17:25 2014

    Beginning log switch checkpoint up to RBA [0xd9.2.10], SNA: 12670277595726 *.

    Thread 1 Advanced to record the sequence 217 (switch LGWR)

    Currently journal # 1, seq # 217 mem # 0: /u01/app/oracle/oradata/TEST/onlinelog/o1_mf_1_938s3lbv_.log

    Currently journal # 1, seq # 217 mem # 1: /u01/app/oracle/fast_recovery_area/TEST/onlinelog/o1_mf_1_938s3nmc_.log

    Wed Apr 16 22:17:36 2014

    Thread 1 cannot allot of new newspapers, sequence 218

    Checkpoint is not complete

    Currently journal # 1, seq # 217 mem # 0: /u01/app/oracle/oradata/TEST/onlinelog/o1_mf_1_938s3lbv_.log

    Currently journal # 1, seq # 217 mem # 1: /u01/app/oracle/fast_recovery_area/TEST/onlinelog/o1_mf_1_938s3nmc_.log

    Wed Apr 16 22:17:40 2014

    Checkpoint completed up to [0xd8.2.10], YVERT RBA: 12670277595657 +++

    Beginning log switch checkpoint up to RBA [0xda.2.10], SNA: 12670277596242 *.

    Thread 1 Advanced to record the sequence of 218 (switch LGWR)

    Notice how we have lines (marked *) what does "beginning log switch checkpoint."

    However, note that control points are not "emergency" until we have a problem with "checkpoint is not complete", and the first control point in the bucket fill in only some time after the second control point in the dumpster.  Log switch checkpoint he still 'place' (or, at least, its necessity is noticed) - but it is not the urgent case where it used to be in early versions of Oracle.

    Concerning

    Jonathan Lewis

  • Cannot archive archive a redo log because the output device is full

    Dear Experts,

    When I'm connected to the MA of my DB 11 g R2. I found the following:

    - State of the database Instance : down
    - Details
    The instance has been terminated by a database due to a deadly internal disease process, or a critical background process was killed by the user.
    Archive is unable to archive a redo log because the output device is full or inaccessible.
    - Connection of the agent to the Instance
    Status: Failed
    Details: ORA-01034: ORACLE not available

    Thank you in advance for any help you provide

    Best regards

    you do not have enough space on your disk,

    You must remove all archivelog/backup obsolete/expired:
    . make a cross-checking first:
    RMAN > obsolete report.
    RMAN > remove obsolete;
    RMAN > list expired backup;
    RMAN > list expired archivelog all;
    RMAN > delete expired backup;
    RMAN > delete all. expired archivelog
    or
    increase the space in your disk.

    To know how many days/backups are in you, the RETENTION POLICY, checking:
    RMAN > show all.

    You can change with the least number of days for example.

    HTH

  • Clearing Online redo logs on the physical standby of the target

    Hello

    Version 11202.

    Primer on computer a.
    Watch on computer b.

    Later note: 11.2 Data Guard physical standby failover best practices using SQL * Plus [1304939.1 ID].
     
    Online redo logs on the target physical standby need to be cleared before that standby database can become a primary database. 
    And later:
     
    On the target physical standby run the following query to determine if the online redo logs have not been cleared... 
    (1) I did not understand where I have to run the query. On computer A, computer B or?
    2)
     
             Online redo logs on the target physical standby need to be CLEARED 
            
    Which means that it must be DELETED? Does transaction could be lost by running the command of compensation?

    Thank you

    Hello again;

    I would NOT use a script. The reason is that you are not sure what "switchover_status" will be returned by the database of v$.

    Oracle a broker data available for a quick pass, but I'm from the old school and I want to see how thing works in SQL for awhile first.

    Best regards

    mseberg

  • Bandwidth of the network for the redo log switches

    Hi all

    I'm using Oracle 11 g R2, I want to clarify the necessary bandwidth of my primary site to the secondary site (using Oracle Data guard)

    My database is an Oracle RAC 2 nodes each node has a log size to 2 GB again. listen to this file in operation every 5 minutes. It means that I'm spending a 2 GB redo log file every 5 minutes from each node. This means that 4 GB in total every 5 minutes.

    I'm planing to set up Oracle Dataguard in maximum performance mode. Is that mean I need a link between the primary and secondary site that is able to transfer files of 2 GB + 2 GB every 5 minutes?

    I do the calculation below and need your advice:

    4 GB / (5x60s) = 13981 KByte/s

    This means that I need to link to 13981x8bites/s = 111, 848Kbit/s?

    Is the above correct? your advice please

    Kind regards

    Please check below two

    http://www.Oracle.com/au/products/database/MAA-WP-10gR2-dataguardnetworkbestpr-134557.PDF

    How to calculate bandwidth transfer network required recovery in care of data (Doc ID 736755.1)

  • datafiles and redo logs on the same drive

    Hi guys,.

    http://docs.Oracle.com/CD/E11882_01/server.112/e25494/onlineredo002.htm#i1306224

    >

    Data files should also be placed on different disks of redo log files to reduce contention in writing of the data blocks and redo records.
    >

    I really think if he actually any challenge when the first of all oracle can only writes files 1 redo log at a time.
    http://docs.Oracle.com/CD/E11882_01/server.112/e25494/onlineredo001.htm
    >
    Database only Oracle uses redo log files at the same time to store redo records written since the restore log buffer by progression. Log roll forward LGWR is preparing actively to is called the log during recovery.
    >

    If the process flow, I got after reading the chapters is

    When LGWR fills a log file of redo (roll forward records) then there will be a log switch + control points where writing for data blocks occur. There seems to be a flow series rather than a simultaneous sort of flow. So I don't really understand it when he speaks of contention will take place when records of data and files written redo which is the redo log file are on the same disks.

    Just to confirm with you guys, whenever there is a switch of newspaper one point of control occurs too right.
    http://docs.Oracle.com/CD/E11882_01/server.112/e25494/onlineredo002.htm? You can search checkpoint to access the section she made mention of this documentation.

    Think about this:

    Restore means to keep the information around in case you need to re-do (recovery or standby mode, i.e. the continuous collection). Updates as you do in a broad sense will be so potentially being twice, once for the data files and once to do it again. I'm sure you understand that the data is written in memory, and then later to the data by the writer of the db files, while there could be many more group of writing data, or they may even be long delayed. In addition, the data files are read at random, so you can't really think again in series compared to the data. Do it again is series, archiving is set, but the data is random, and random how depends on how your system is used.

    So that means all the type of reads and writes per redo and archive is fundamentally different from data and cancel. In the first case, you want to be able to breath out as I/O that you can, for the latter, you want to be able to randomly reading or writing at different times, with Oracle being smart enough to do a bit of it in memory and optimistic enough to make assumptions on when to do things and quite lazy for not doing everything right. Roll forward is critical.

    A while ago, someone pointed out that the e/s modern buffered in memory, do not really worry about this, because all the work required to set up and maintain it after spending records, fees are not much better than striping and mirroring everything (you can google SAMI). This is true to a point, and we can debate endlessly about RAID types and their effects on performance and how their buffering makes [url http://www.baarf.com/] useless BAARF. But the real debate is, where is the point that you should use separate to redo and data devices? In the real world, we often receive a standard hardware configuration, which works very well until it's not. A disk or controller will puff in a RAID-5 can happen "is not" real quick.

    You should probably take two thoughts:

    The docs are pretty General, and some old tips do not apply, may have transformed into myth or perhaps too general to be meaningful.

    There is always something in the db, and the more things underway, less you can make generalizations about the serialization.

  • Standby Redo logs and the Directory Structure in the Backup Site

    Hi guru

    I just want to confirm, I know that if the directory structure is different, I need to talk about these 2 settings in the file pfile

    on the main site:

    DB_CONVERT_DATAFILE = 'sleep', 'primary '.

    LOG_CONVERT_DATAFILE = 'sleep', 'primary '.

    On the secondary Site:

    DB_CONVERT_DATAFILE = 'primary', 'sleep '.

    LOG_CONVERT_DATAFILE = 'primary', 'sleep '.

    But I want to confirm this weather I have to deliver the full path of the directory in the two paramtere above:

    as:

    DB_CONVERT_DATAFILE = ' / u01/oracle/app/oracle/oradata/sleep ', ' / u01/oracle/app/oracle/oradata/primary.

    LOG_CONVERT_DATAFILE = ' / u01/oracle/app/oracle/oradata/sleep ', ' / u01/oracle/app/oracle/oradata/primary.

    Second Confusion: -.

    After that transfer standby Redo logs creates primary and made to sleep on the foregoing, mentioned the directory structure and after restoring the alongwith primary db backup, that ensures the control file will not affect the journal of physics again placed watch on the above mentioned location.

    Thanks in advance for your help

    vk82 wrote:

    In fact, I create the day before by using the RMAN Duplicate command. but where I am confused is the point if I transfer the backup that i taken the pri on the C:\backup_files path. and after that, I transfer the backups to the waiting on C:\backup_files. After that when I restore it will create datafile and other stuff in another directory I mentioned using DB_FILE_NAME_CONVERT as LOG_FILE_NAME_CONVERT. I think Yes but need your advice during the same period.

    Hello

    Yes, files should be created under the directory mentioned in the path of the parameters 'db_file_name_convert' and 'log_file_name_convert.

    Kind regards
    Shivananda

  • redo log switch

    11 GR 2
    Discovered newspaper spent all the minutes (2, 3 minutes) at peak periods. I'll make the recommendation for the largest newspapers of recovery such as switches will be every 20 to 30 minutes.
    Wanted to know what would be the negative side of the great redo logs?

    You have the active setting FAST_START_MTTR_TARGET? If you have it, you can use the Redo size for the log file Advisor who can tell you the right size for them for your redo log files. In addition, you can set parameter archive_lag_target while fixing a large size for her redo log files.

    Aman...

  • How to replace the members of the Group missing redo log after the failure of a support

    Hello

    I'm testing a disk failure scenario.  I had recovery multiplexes on three sites (Windows drive letters) logs; one of the three locations also contained the data files.  I deleted the Windows player that contained data files and saw a restoring with RMAN.   All is well except that redo log members who are found on the occult Windows disk are of course still missing; the database opens, etc., but puts messages like this in the ALERT.log:

    ORA-00321: log thread 2 1, cannot update log file header
    ORA-00312: wire 2 1 online journal: ' F:\APP\NGORA12CR2SRV\ORADATA\ORATEST5\ONLINELOG\O1_MF_2_BW4B5MY0_. JOURNAL"


    What is the 'right' way to replace missing limbs?  Just get all the new groups of newspapers with new members on all three discs?  Carefully copy the LOG files representing missing members of one of the other sites?  I add new members at this location and remove existing members (missing)?


    Is how important it?


    Thank you!


    Martin

    See 'Recover after losing a member of a group of Redo Logs online multiplexed'

    at http://docs.oracle.com/cd/E11882_01/backup.112/e10642/osadvsce.htm#sthref2205

    Hemant K Collette

  • Can redundancy of the OCR - we have OCR mirror with Redo logs in the diskgroup REDO01?

    Hello

    I have 2 node Oracle 11 GR 2 (11.2.0.3) RAC on AIX 6.1. Currently there is only an OCR and a VD residing in OCRVD diskgroup what redundancy as external.

    # File group universal name Id disk STATE file
    -- ----- ----------------- --------- ---------
    1. online a27d6a5df2db4f31bf0c9c9d31870b1f (/ dev/rhdisk11) [OCRVD]
    Located at 1 vote (s) disk.


    $ ocrcheck
    Status of the Oracle Cluster registry is:
    Version: 3
    Total space (in kilobytes): 262120
    Space (in kilobytes) used: 3152
    Amount of available space (KB): 258968
    ID: 899322289
    File name of device /: + OCRVD
    Verifying the integrity of file/device succeeded

    File/device not configured

    File/device not configured

    File/device not configured

    File/device not configured

    Checking cluster registry integrity succeeded

    Logical corruption control bypassed due to the non-privilegie user

    OCRVD Diskgroup is created using 5-disc size 1 G with RAID 1.

    1. is it possible that I have OCR and VD multiplex in the same diskgroup?
    2. If it is not possible, then I would add OCR and VD in other starts (Eg: REDO01 and REDO02 - so much for Redo logs)?

    Published by: Azam on 20 may 2013 14:46

    Azam says:
    Hello

    I have 2 node Oracle 11 GR 2 (11.2.0.3) RAC on AIX 6.1. Currently there is only an OCR and a VD residing in OCRVD diskgroup what redundancy as external.

    # File group universal name Id disk STATE file
    -- ----- ----------------- --------- ---------
    1. online a27d6a5df2db4f31bf0c9c9d31870b1f (/ dev/rhdisk11) [OCRVD]
    Located at 1 vote (s) disk.

    $ ocrcheck
    Status of the Oracle Cluster registry is:
    Version: 3
    Total space (in kilobytes): 262120
    Space (in kilobytes) used: 3152
    Amount of available space (KB): 258968
    ID: 899322289
    File name of device /: + OCRVD
    Verifying the integrity of file/device succeeded

    File/device not configured

    File/device not configured

    File/device not configured

    File/device not configured

    Checking cluster registry integrity succeeded

    Logical corruption control bypassed due to the non-privilegie user

    OCRVD Diskgroup is created using 5-disc size 1 G with RAID 1.

    1. is it possible that I have OCR and VD multiplex in the same diskgroup?

    No, this isn't posible

    2. If it is not possible, then I would add OCR and VD in other starts (Eg: REDO01 and REDO02 - so much for Redo logs)?

    Yes you can add other starts. As root, run the following command to add a location of OCR or Oracle ASM

    ocrconfig -add +REDO01
    ocrconfig -add +REDO02
    

    As you know on level between discs ASM ASM mirroring that is contains ASM DISKGROUP.

    Concerning
    Mr. Mahir Quluzade

  • Adding STAND OF REDO LOG in the primary side...

    Hi all

    I set STANDBY_FILE_MANAGEMENT = AUTO side pending and LOG_FILE_NAME_CONVERT points to a location that exists at the level of the BONE...
    When I added a data file to an existing table in the primary circuit and made a journal on the same order. The obtained additional data file
    relflected next to the day BEFORE...

    But when I try to add that a log file pending (new group) does not have added...

    Please help me on this...


    For this I need to perform the steps below.
    1. Add a redolog watches over the primary side...
    2. then cancel the recovery managed by the side process ensures...
    3 - Add the redolog ensures even (same name and size) on the side pending
    4 put the watch into recovery mode...

    Thank you

    STANDBY_FILE_MANAGEMENT = AUTO for data files is...

    See Doc...
    http://docs.Oracle.com/CD/B14117_01/server.101/b10755/initparams206.htm

  • Copy the datafiles/redo logs from the Production Server to QA Server

    I need to update the QA database on server QA with the Production database on the Production Server. I intend to copy logs files from data/recovery of the Production database to replace the QA database datafiles/recovery logs. Should I stop the Production database before you start copying? If this is not the case, what happens if a Production database data file has been copied to QA and then this Production obtained data file updated before other newspapers of data/recovery files have all been copied to QA?

    Thanks for any help.

    881656 wrote:
    Should I stop the Production database before you start copying? If this is not the case, what happens if a Production database data file has been copied to QA and then this Production obtained data file updated before other newspapers of data/recovery files have all been copied to QA?

    If you use this technique without closing the database, copied datafile/recovery logs will be incompatible.

    You can use the RMAN utility to take backup of the production database and restore it on server QA.

    If the database size is small (about 10-20 gb) you can use EXPDP/IMPDB.

  • Preferred method for shipping redo logs in the standby mode: archive or LGWR

    DB version: 11.2
    OPERATING SYSTEM: AIX


    Which is the preferred method for shipping again connects standby: archive or LGWR?

    Citizen_2 wrote:
    DB version: 11.2
    OPERATING SYSTEM: AIX

    Which is the preferred method for shipping again connects standby: archive or LGWR?

    LGWR is recommended where repeat applies will be interms of roll forward. If in the case of all redo currently lost, you have very less time loss of data. Real-time here APPLY.
    In the case of ARCH, transferred with regard to Archives and applied.

    See the http://docs.oracle.com/cd/B19306_01/server.102/b14239/log_transport.htm documentation
    Thank you.

  • Switch redo logs every hour to provide point-in-time?

    Hi all

    Oracle 11G on Windows 2008 R2

    I I learn the details of the backup/restore management and want to be able to provide a good point-in-time capacity in the event of a disaster. It is advisable to plan a hourly job making ALTER SYSTEM SWITCH LOGFILE;? This causes the redo log archiving every hour so if I need to restore, I know that I can restore at least the previous hour (using newspapers to archive) and before?

    Thanks in advance

    From time to time, this may come as a result of the excessive workload. Having the EA edition, having defined FAST_START_MTTR_TARGET? If Yes, you can take use Advisor Redo Log suggest the adequate size of logs again for you. A safer side, you can also increase the size of f they are greater as 1gig and order the rate switching by using the parameter Archive_lag_Target.

    HTH

    Aman...

  • How to find the configuration state of REDO logs files?

    I'm trying to move the redo log files.
    Before that, I want to know if it is set as a duplex or any other configuration.

    How to find the REDO log configurations.

    Oracle 10g R2

    Thank you
    Smith

    Example:

    Not in duplex redo log - if the number of members is 1, it is not duplex:

    SQL> select group#, members from v$log;
    
        GROUP#    MEMBERS
    ---------- ----------
             1          1
             2          1
             3          1
    
    SQL> select group#, member from v$logfile;
    
        GROUP# MEMBER
    ---------- --------------------------------------------------
             1 /u01/app/oracle/oradata/db1/redo01.log
             2 /u01/app/oracle/oradata/db1/redo02.log
             3 /u01/app/oracle/oradata/db1/redo03.log
    
    SQL> alter database add logfile member '/u01/app/oracle/oradata/db1/redo01_2.log' to group 1;
    
    Database altered.
    
    SQL> alter database add logfile member '/u01/app/oracle/oradata/db1/redo02_2.log' to group 2;
    
    Database altered.
    
    SQL> alter database add logfile member '/u01/app/oracle/oradata/db1/redo03_2.log' to group 3;
    
    Database altered.
    

    Duplex redolog

    SQL> select group#, members from v$log;
    
        GROUP#    MEMBERS
    ---------- ----------
             1          2
             2          2
             3          2
    
    SQL> select group#, member from v$logfile;
    
        GROUP# MEMBER
    ---------- --------------------------------------------------
             1 /u01/app/oracle/oradata/db1/redo01.log
             2 /u01/app/oracle/oradata/db1/redo02.log
             3 /u01/app/oracle/oradata/db1/redo03.log
             1 /u01/app/oracle/oradata/db1/redo01_2.log
             2 /u01/app/oracle/oradata/db1/redo02_2.log
             3 /u01/app/oracle/oradata/db1/redo03_2.log
    

Maybe you are looking for

  • Error recovery system...

    Hello.. I use Hp pavilion g6 2020se and recently upgraded to win8 win 7 Home Basic. Files Windows are under C: drive that is 100 GB (resized earlier), logical partition. Hard disk also contains the recovery partition. I want to downgrade to win 7. I

  • How to install updates that does not have install for Windows vista

    Update KB 3178034, KB3167679, KB3177725, KB3175443, tried to install, but failed. It is a windows Vista Home Premium 64-bit operating system. These updates can be installed and how? Windows Vista is still a usable operating system?  I'd appreciate an

  • AnyConnect configuration using IPSec

    I have configured our ASA 8.4 (7) running for the client AnyConnect (using IPSec). It prompted me to create an identity certificate when you run the VPN Wizard, I did. We use AAA to authenticate, so I not create a CA. It must in any case for AnyConne

  • How to merge video projects into Adobe premiere elements 14

    I am running Adobe premiere elements 14 on an imac and I built two projects that I am trying to merge into a single project. Is it possible to do without exporting to a video file, so I can still change the two projects into one project? Thank you in

  • Long data type value does not block PL/SQL

    HelloI wrote a stored procedure, the procedure must return high_value of all_tab_partitions table. that contains the value, but giving as no data found. Here is the code used to run.CREATE or REPLACE PROCEDURE PR_HDR_FNL_CLAIM_LINE isL_DATA varchar2