DBAsupport.com Forums - Powered by vBulletin
Results 1 to 5 of 5

Thread: very high "redo buffer allocation retries " value

  1. #1
    Join Date
    Dec 2001
    Posts
    120

    very high "redo buffer allocation retries " value

    hi everybody,

    thanks first of all for your earlier suggestions.

    I am facing problems regarding the speed of inserts into my oracle 9i database. The oracle software & the WIN-2000 server software are on a separate physical disk(C Drive). Rest of the files viz: the redo logs,controll files , user tablespace data files, index tablespace files , system tablespace data files, undofiles are all present on another physical hard disk(D Drive). I am doubting some kind of contention issues because of all these files present in one disk(D Drive). Plz tell me about which of these files should be separated and should not be kept together on the same disk.

    should the redo log files be moved to the C Drive which presently has only the oracle software and windows running in it?

    I am providing 3 queries which shows that there is a wait for free log buffers and this could be the actual cause of poor inserts. The log buffer size is 10MB and there are 4 online redo log files of 100MB each. The RAM size is 1GB and sga_max_size is 750MB.

    --------------------------------------------------------------------
    1) select name,value from v$sysstat where name = 'redo buffer allocation retries';
    Result:- redo buffer allocation retries -> 107

    2) select name,value from v$sysstat where name = 'redo log space requests';
    Result:- redo log space requests -> 133


    3) select r.value "Retries", e.value "Entries" , r.value/e.value*100 "Percentage" from v$sysstat r, v$sysstat e
    where r.name = 'redo buffer allocation retries' and e.name = 'redo entries';
    Result:- Retries = 107, Entries = 16249464 , percentage = 0.00065848325827855 ;

    ----------------------------------------------------------------------

    pls give your suggestions on proper grouping of oracle data files and there placement on a disk and how to decrease the "redo buffer allocation retries" problem...


    thanking all of you

    Parijat Paul

    redo buffer allocation retries

  2. #2
    Join Date
    Sep 2000
    Location
    Chennai, India
    Posts
    865

    Re: very high "redo buffer allocation retries " value

    Originally posted by parijat67
    I am facing problems regarding the speed of inserts into my oracle 9i database.
    If are doing large bulk-inserts, only during a period of time, consider dropping indexes before the bulk insert and then rebuild indexes in parallel after the bulk insert.

    HTH.

  3. #3
    Join Date
    Nov 2002
    Location
    Geneva Switzerland
    Posts
    3,142

    Re: very high "redo buffer allocation retries " value

    Originally posted by parijat67
    should the redo log files be moved to the C Drive which presently has only the oracle software and windows running in it?
    This may be better, but you will probably have some contention with the OS paging. I found this a "problem" for log switches (I don't know why) - you have big log files, my guess is that you would win.

  4. #4
    Join Date
    Dec 2001
    Posts
    120
    i have changed the server hardware to tackle the problem of slow writing to the redo log files and the corresponding log buffer wait. Earlier it was a single CPU based server machine and now we have put double CPU and the RAM has been increased to 2Gb from 1 Gb. Oracle has been alloted 1GB of RAM. The new memory allocation is as follows:-
    SGA_MAX_SIZE = 1GB
    DB_CACHE_SIZE = 400Mb
    SHARED_POOL_SIZE= 400Mb
    LARGE_POOL_SIZE = 30Mb
    JAVA_POOL_SIZE= 30Mb
    LOG_BUFFER= 15Mb (EARLIER THIS WAS 10M)
    PGA_AGGREGATE_TARGET= 50Mb

    all the oracle datafiles are at e: drive whereas the oracle software is installed at c: drive. both these drives are logical drives on the same physical drive.

    After these modifications we did an database import to import an user data. After the completion of the imports i found that still the value for "redo buffer allocation retries" is more than 0 and "redo log space requests" is also greater than 0, its 10 & 11 respectively right now. The logging for user data as well for indexes has been switched off at the tablespace level and the database is running in "noarchive mode". What should we conclude from these statistics? oracle documents mention that these values should be 0 all the time or else the log_buffer should be tuned. How should be tune the log_buffer?? pls send your suggestions on this regard.

    thanks

    parijat

  5. #5
    Join Date
    Jul 2001
    Location
    Slovenia
    Posts
    422
    This does not necessarily mean you have log contention, since both values are quite low comparing to redo entries.
    Check log buffer space waits to see how much processes wait because of log contention.
    Tomaž
    "A common mistake that people make when trying to design something completely
    foolproof is to underestimate the ingenuity of complete fools" - Douglas Adams

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  


Click Here to Expand Forum to Full Width