IT32869: PRIMARY HUNG WHEN HADR_FLAGS= STANDBY_RECV_BLOCKED SSL_PROTOCOL

Db2 Version 11.1 Mod 4 Fix Pack 6 for Linux, UNIX, and Windows
Db2 Version 11.1 Mod 4 Fix Pack 6 for Linux, UNIX, and Windows

APAR status

• Closed as program error.

Error description

• When using SSL the standby may get into a state
  where it becomes blocked with:
    HADR_FLAGS = STANDBY_RECV_BLOCKED SSL_PROTOCOL
  
  Causing the primary to hang and get stuck in:
  
  0x09000000001180B8 thread_wait + 0x98
  0x0900000079BA2A34 sqloWaitThreshold + 0x994
  0x090000007FA4583C
  sqlpgWaitForLrecBufferLimit__FP14sqlpMasterDbcbCUl + 0x1DC
  0x0900000078E5437C
  sqlpWriteToLog__FP8sqeAgentP11SQLP_TENTRYUlN23P14SQLP_LREC_PARTP
  9SQLP_LSN8PUl + 0x19DC
  0x0900000078E59EB0
  sqlpWriteLR__FP8sqeAgentUlN42P14SQLP_LREC_PARTT2P9SQLP_LSN8PUl +
  0xCF0
  0x090000007B2E6848 IPRA.$sqldWriteDeleteLR__FP13SQLD_DFM_WORKiUl
  + 0x1E8
  0x090000007B2E5500
  sqldLogAndDeleteRecord__FP13SQLD_DFM_WORKiT2Ul + 0x500
  0x090000007B2EFD44 sqldRemoveRow__FP13SQLD_DFM_WORKi + 0x54C4
  0x090000007B2F9268 sqldDeleteRow__FP13SQLD_DFM_WORKiUl + 0xDE8
  0x090000007C5DFFA8 sqldRowDelete__FP8sqeAgentP8SQLD_CCBUlPPv +
  0x8E8
  0x090000007EDE3988 sqlridel__FP8sqlrr_cb + 0x408
  
  This may be due to the following reasons:
  1) using ASYNC, performing large uow (ie. each IO by loggw is
  writing the entire log buffer)
  2)standby receive buffer default is 2*LOGBUFSZ, but standby
  cannot consume the received data yet, until next communication
  with primary
  this is in case primary crashed and the received log data was
  not written to disk on primary(primary overlaps network transfer
  with its own writing to disk, so such possibility exists).
  3)next IO is also entire buffer, and data transferred was
  encrypted(because SSL_PROTOCOL), and we need to decrypt the data
  using the same buffer, which must need a bit extra to facilitate
  the decryption.
  
  We should change our default STANDBY_RECV_BUF_SIZE: ie. standby
  receive buffer should be a bit more than 2*LOGBUFSZ.
  

Local fix

• Making the DB2_HADR_BUF_SIZE a bit more than 2*LOGBUFSZ.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * THE HADR_FLAGS MAY SHOW STANDBY_RECV_BLOCKED SSL_PROTOCOL    *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * See Error Description                                        *
  ****************************************************************
  * RECOMMENDATION:                                              *
  * First fixed in v111m4fp6                                     *
  ****************************************************************
  

Problem conclusion

• First fixed in v111m4fp6
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  DB2 FOR LUW

• Fixed component ID

  DB2FORLUW

Applicable component levels

• RB10 PSY

     UP