2841452 - RepServer hangs while trying to resolve an error on a database connection - SAP SRS

• SRS gets hung / becomes unresponsive after reporting the following message:

A grouped transaction of xxx individual transactions has failed in database 'DS_WS.mydb'. Each transaction in the group will be executed individually.

• Before this happens, tons of foreign key constraint violation errors are detected on this database connection which can sometimes bring the DSI EXEC thread down after 400 retries but not always (not if the issue is resolved within 400 retries), for example:

WARNING #5091 DSI EXEC(390(1) DS_WS.mydb) - neric/dsi/dsiqmint.c(5345)
A transaction for database 'DS_WS.mydb' failed. It will be retried 400 times. The data server error received (#546) is mapped to RETRY_LOG or RETRY_STOP.
ERROR #1028 DSI EXEC(390(1) DS_WS.mydb) - neric/dsi/dsiqmint.c(5165)
Message from server: Message: 546, State 1, Severity 16 -- 'Foreign key constraint violation occurred, dbname = 'mydb', table name = 'dbo.mytable', constraint name = 'mytable_fk'.'.
Message from server: Message: 3621, State 0, Severity 10 -- 'Command has been aborted.'.
THREAD FATAL ERROR #5049 DSI EXEC(390(1) DS_WS.mydb) - neric/dsi/dsiqmint.c(5178)
The DSI thread for database 'DS_WS.mydb' is being shutdown. DSI received data server error #546 which is mapped to STOP_REPLICATION. See logged data server errors for more information. The data server error was caused by output command #1 mapped from input command #2 of the failed transaction.
THREAD FATAL ERROR #5273 DSI EXEC(390(1) DS_WS.mydb) - neric/dsi/dsiqmint.c(5206)
To write the failed transaction into log, please execute 'sysadmin log_first_tran, DS_WS, mydb'. Please analyze the transaction and provide an appropriate fix based on your analysis, then resume the connection.
The DSI thread for database 'DS_WS.mydb' is shutdown.
The DSI thread for database 'DS_WS.mydb' is started.
A grouped transaction of 100 individual transactions has failed in database 'DS_WS.mydb'. Each transaction in the group will be executed individually.

• The SRS process on the host has to be killed from the OS side.

• On the first SRS restart with the diag binary, SRS reports Signal 11 in mem_free() for the impacted database connection and crashes  with a core dump right after the DSI EXEC thread resumes work:

A grouped transaction of 100 individual transactions has failed in database 'DS_WS.mydb'. Each transaction in the group will be executed individually.
(159): Thread DSI EXEC(390(1) DS_WS.mydb) infected with signal 11.
(159): Dumping context structure:
(159): pc(iar) = 2446d8p
(159): Registers: g0=2447c8, g1=638ccd50, g2=102b59c0, g3=6e652c20, g4=d0000, g5=8, g6=42e0, g7=0, g8=15f1002f, g915f1002f, g10=96e0, g11=0, g12=241038, g13=638d7800, g14=0, g15=0
(159): Registers: g16=0, g17=0, g18=0, g19=0, g20=0, g210, g22=0, g23=0, g24=0, g25=0, g26=0, g27=0, g28=0, g29=0, g30=103c1f40, g31=bea890
....
(159): Dumping memory trace.
(159): ***************STACK TRACE***************
(159): *****thread DSI EXEC(390(1) DS_WS.mydb)*******
(159): pc: 0x0000000000248b80 + 00000000000001 fc, .exc__appDumpStack ()
(159): pc: 0x0000000000524f80 + 00000000000004 b8, .exc_terminate ()
(159): pc: 0x0000000000000ba0 + 0000000000000934 , .bsd_catch_signal ()
(159): pc: 0x0000000000243de0 + 00000000000008 f8, .mem_free ()
(159): pc: 0x0000000000981a60 + 0000000000002300 , .dsie__wrap ()
(159): pc: 0x000000000024dda0 + 00000000000002 c8, .srv__start_function ()
(159): pc: 0x0000000000516e14, unknown dynamic function
(159): *****End of stack trace.****
(159): Internal error. Attempting to produce a core file

• The next SRS restart is successful.  Normal replication resumes with no losses.

SAP Replication Server (SRS) 16.0

cross-database, cross, referential, integrity, dsie__wrap, rv__start_function, SEGV, segmentation, storage, access, fault, segfault, stacktrace, stack , KBA , BC-SYB-REP , Sybase Replication Server (standalone) , Problem