Do you see a ‘Congested’ State for HADR while Performing Reorgs?

You may also like...

6 Responses

  1. memet_tiike says:

    Hi Ember,
    I am reading about HADR_SPOOL_LIMIT parameter in information center and blogs and I am confused about this parameter with SYNC HADR mode. HADR_SPOOL_LIMIT enables primary to continue transaction processing without waiting for standby to log replay. On the other hand, for SYNC HADR mode, primary waits for standby to write the transfered log to its log file. I understand that HADR_SPOOL_LIMIT has no use for SYNC mode? Am I missing anything?

  2. Divakar says:

    Hi Ember,

    The chances of DB going in the congested state are more in case on Online Index reorg rather than Table reorg. Do you know this is so.

    • Ember Crooks says:

      If you do only online reorgs, you must perform index reorgs in addition to the table reorgs. I don’t think I’ve tested it to understand which causes more impact.

  3. DB2Jabber says:

    To make sure I understand completely, you are suggesting to use the DB CFG parameter of HADR_SPOOL_LIMIT and to not use the DB2 registry parameter DB2_HADR_BUF_SZ, but i must ask about a couple settings left-over from DB2 9.5, we still have 2 DB2 Registry parameters in use DB2_HADR_SORCVBUF and DB2_HADR_SOSNDBUF, should we discontinue it’s use. Thank you for your time and valuable input.

    • Ember Crooks says:

      DB2_HADR_SORCVBUF and DB2_HADR_SOSNDBUF are still valid, but they relate to the buffer size on each side. they have nothing to do with spooling logs, and are often less than 512K.

  4. E. Crider says:

    Note: There is a bug in DB2 10.1 with HADR_SPOOL_LIMIT DB Config when using values other than 0. The issue can result in HADR not being able to start cleanly or the standby crashing. The issue was first fixed in DB2 10.1 FP 6. See details on tech note for “APAR IT13427: STANDBY MIGHT CRASH WITH SPOOLING ENABLED ONLY IN V10.1”

Leave a Reply

Your email address will not be published.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