2451995 - A secondary server got crashed with "blockmap userlocked during flush".

1. A secondary server got crashed while processing ev_begintxn.
0000324777 Connect: SA connHandle: 1000210897 SA connID: 16 IQ connID: 0000324777 User: dbo
0000324777 ev_begintxn started
0000324777 Cmt 26917319
..................
0000324777 TxnWt 26917413 0 26918255
0000324777 blockmap userlocked during flush; _blockmap @ 0x7fe34526b4d8
0000324777 Blockmap Member Variables for 0x7fe34526b4d8
0000324777 From dblib/db_table.cxx:3052
0000324777 _nextPartitionForHelperFlush = 0
..................
0000324777 **************************************************
0000324777 *** SAP IQ Abort:
0000324777 *** From: slib/s_bufman.cxx:5460
0000324777 *** PID: 12042
0000324777 *** Message: blockmap userlocked during flush
0000324777 *** Thread: 140657276188416 (TID: 23)
0000324777 **************************************************

===== Thread Number 140657276188416 (IQ connID: 0000324777) =====

0000324777 pc: 0x7fed4d4207b2 pcstkwalk(..........
0000324777 pc: 0x7fed4d420e21 ucstkgentrace(..........
0000324777 pc: 0x7fed4d4221f9 DumpAllThreads(..........
0000324777 pc: 0x7fed4d19e71a hos_ABORT(..........
0000324777 pc: 0x7fed4d5a86ac s_bufman::PrivateFlush(..........
0000324777 pc: 0x7fed4d5a881b s_bufman::FlushToDisk(..........
0000324777 pc: 0x7fed4f1ff93e st_txncb::FlushCoalesce(..........
0000324777 pc: 0x7fed4f20b13d st_txncb::FlushForCommit()+..........
0000324777 pc: 0x7fed4f226a17 st_txnMgr::FlushForCommit()+..........
0000324777 pc: 0x7fed4deef8d2 st_command::DoCmdThroughResourceGate()+..........
0000324777 pc: 0x7fed4df436f3 st_iqtxn::FlushForCommit(..........
0000324777 pc: 0x7fed4ccb4c35 UIQTxn_FlushForCommit+..........
0000324777 pc: 0x7fed4df48346 st_SAIQInterfaceInfo::callFunction()+..........

2. This problem is identical with CR 789421 which was fixed in IQ 16.0 SP10 PL11.

SP10.11 789421 DML Statements on a secondary writer node may cause intermittent crashes on that writer at s_bufman::PrivateFlush, hs_dpInsert::FinishedInsert(),
st_iqtxn::PostCommitTxn when tlv log replay occurs on that writer. This has been fixed.

In that CR, the IQ secondary server got crashed while loading data.

void hos_ABORT(..........
SigHndlr+0x41(..........
_fini+0x173fc216(..........
_fini+0x173f0ae2(..........
_fini+0x173f0d0e(..........
_fini+0x12830cf(..........
void s_bufman::PrivateFlush(..........
void s_bufman::FlushToDisk(..........
void st_txncb::FlushCoalesce(..........
void st_txncb::FlushForCommit()+..........
void st_txnMgr::FlushForCommit()+..........
void st_command::DoCmdThroughResourceGate()+..........
unsigned st_iqtxn::FlushForCommit(..........
UIQTxn_FlushForCommit+..........
void st_SAIQInterfaceInfo::callFunction()+..........

The whole pattern to the crash is identical with CR 796339, which is duplicate with CR 789421.

3. The other secondary servers were working fine when this secondary server got crashed.

4. The customer ran DBCC against the database with allocation mode and no error detected.

================================================== ==============================
DBCC Allocation Mode Report
================================================== ==============================
DBCC Status No Errors Detected

SAP IQ 16.0 SP11 PL04
Multiplex

SAP IQ, IQ, Abort, Crash, 806919, CR806919, CR 806919, blockmap userlocked during flush, s_bufman, PrivateFlush, s_bufman::PrivateFlush, FlushToDisk, s_bufman::FlushToDisk, slib/s_bufman , KBA , BC-SYB-IQ , Sybase IQ , Problem