Results 1 to 9 of 9
  1. #1
    Join Date
    Mar 2003
    Location
    Singapore
    Posts
    200

    Unanswered: Warning: active log held by dirty pages.

    Hi ,

    Recently i have been getting this message in my db2diag.log.
    This occurs once a week at a paricular time(12:30 midnight) ,not on a particular day.I checked ,at this particular time there is no particular scheduled actions as backups or so.

    I tried adjusting the values many times,still now i am recieving the error.

    Below i have extracted a portion of the log file for your reference.

    ----
    2003-07-02-04.00.32.734000 InstanceB2 Node:000
    PID:1288(db2syscs.exe) TID:920 Appid:none
    data_protection sqlpgrsp Probe:20

    Warning: active log held by dirty pages.
    Decrease softmax and/or increase num_iocleaners.

    2003-07-06-04.00.35.734000 InstanceB2 Node:000
    PID:1288(db2syscs.exe) TID:920 Appid:none
    data_protection sqlpgrsp Probe:20

    Warning: active log held by dirty pages.
    Decrease softmax and/or increase num_iocleaners.

    ---------

    Why is this occuring and what can i do next.

    Thanks in advance

    micky

  2. #2
    Join Date
    May 2003
    Location
    USA
    Posts
    5,737
    I would do as they suggest:

    Decrease softmax and/or increase num_iocleaners.

    Also, you might need to look at certain applications and do more frequent commits.

  3. #3
    Join Date
    Mar 2003
    Location
    Singapore
    Posts
    200
    I also tried decreasing and increasing the parameters,but did not get any result.

    Also my auto commit is set on,i hope it will commit after each transaction.

    Should i increase the size of the log files will this help.

    mickykt

  4. #4
    Join Date
    Mar 2003
    Posts
    343
    Did you deactivate and activate the database after making the db cfg changes?

    Do the applications commit frequently?

    How does the issue affect system operation?

  5. #5
    Join Date
    Mar 2003
    Location
    Singapore
    Posts
    200
    Hi,

    I don't feel any effect on the entire system using the database.The Only thing i can see is in the db2 log file an error message.

    Yes i restarted the database after changing the values.

    First i though that any backup schedule was happening at this time because it always happened at 12:30 midnight(not a specific day).But i verified nothing is shceduled at 12:30.

    My concern was since it was holding the log files,is there any effect of increasing the size / or number o log files .

    Pls ask if u need anyspecific file or error.

    Thankx
    mickykt

  6. #6
    Join Date
    Mar 2003
    Location
    Singapore
    Posts
    200
    Hi all,

    I also have a few doubts,

    What is soft max,when i read it stated - This is used to influence the number of logs that need to be recovered following a crash and determine the frequency of soft checkpoints. Range [1-100*logprimary]Percentageof one primary log file.

    So this means if i set softmax=80,with primary log file size 250*4K

    what is the number of log files to be recovered following a crash .

    is it 80*250*4/100=800 logfiles or

    80*250/100=200 logfiles.

    I usually delete all the log files that are obsolete(that before a offline backup).
    Will this create problems?

    How does this parameter also affect the frequency of soft checkpoint?so if i decrease the values the softcheckpoint also decreases,there by holding the logfiles for more time.Is this what i have understood correct or wrong.

    pls help

    mickykt

  7. #7
    Join Date
    Jan 2002
    Location
    Manila, Philippines
    Posts
    71

    parameters ?

    Can you pls post your value for the following DB CFG parameters ?
    LOGBUFSIZ
    BUFFPAGE
    LOGFILSIZ
    LOGPRIMARY
    SOFTMAX

    Thanks !

  8. #8
    Join Date
    Mar 2003
    Location
    Singapore
    Posts
    200
    Hope this is sufficient,

    LOGBUFSIZ=32 * 4K
    BUFFPAGE=40352*4K
    LOGFILSIZ=250*4K
    LOGPRIMARY=6
    SOFTMAX=60
    NUM_IOCLEANER=10

    Thanks for ur concern and help.

    mickykt

  9. #9
    Join Date
    Mar 2003
    Posts
    343
    According to your db cfg, the logfilesize is 1m. To my knowledge, softmax works as

    100% means 1 log file(1MB in your case), 60% means .6 of a log file or (600KB in you case).

    Which means that if the database crashes, db2 will have 600KB of uncommitted to disk changes in the logfile to go through(also called crash recovery) when it comes up.

    In my opinion, the error in the diag is due to a fraction of the logfile being specified for commits to disk. ie, when it the changes in the logfile exceed 600KB, it is trying to commit those transactions but 40% of the logfile has dirty pages. Just my opinion. Try increasing this value to 100 and see if it recurs.

    To understand the impact of this parameter it is necessary to understand how the bufferpool is managed - there is a comprehensive section on it in the Performance manual at
    http://www-3.ibm.com/cgi-bin/db2www/...7d0bufpool.htm

    Take a look at this thread about deleting log files.
    http://dbforums.com/t840601.html
    All of it may not apply to your situation since you seem to be doing offline backups.

    Hope this helps.

Posting Permissions

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