Re: Server Crash due to assertion failure in _bt_check_unique()

Hi All,

I'm getting a server crash in _bt_check_unique() when running the following test-case.

Steps to reproduce the crash:
Step1: Create a test table with primary key and insert some data in it.

create table t1 (a integer primary key, b text);

insert into t1 values (1, 'text1');
insert into t1 values (2, 'text2');
insert into t1 values (3, 'text3');
insert into t1 values (4, 'text4');
insert into t1 values (5, 'text5');

Step2: Start 3 backend sessions and run the following anonymous block in each of them in parallel:

Session 1:

do $$

declare

begin

 insert into t1 values (6, 'text6');

 update t1 set b = 'text66' where a=6;

 perform pg_sleep(7);

 delete from t1 where a=6;

end $$;

Session 2:

do $$

begin

 insert into t1 values (6, 'text6');

 perform pg_sleep('7');

 delete from t1 where a=6;

end $$;

Session 3:

do $$

begin

 insert into t1 values (6, 'text6');

 delete from t1 where a=6;

end $$;

Here is the backtrace for the crash:

#0  0x00007f096019f277 in raise () from /lib64/libc.so.6
#1  0x00007f09601a0968 in abort () from /lib64/libc.so.6
#2  0x0000000000a54296 in ExceptionalCondition (conditionName=0xafdbf8 "!(!_bt_isequal(itupdesc, itup_key, page, offset))",
    errorType=0xafd75a "FailedAssertion", fileName=0xafd81f "nbtinsert.c", lineNumber=386) at assert.c:54
#3  0x0000000000509b0a in _bt_check_unique (rel=0x7f096101a030, insertstate=0x7ffcbd5db9d0, heapRel=0x7f0961017c00, checkUnique=UNIQUE_CHECK_YES,
    is_unique=0x7ffcbd5dba01, speculativeToken=0x7ffcbd5db9c8) at nbtinsert.c:386
#4  0x00000000005096ab in _bt_doinsert (rel=0x7f096101a030, itup=0x27bedc0, checkUnique=UNIQUE_CHECK_YES, heapRel=0x7f0961017c00) at nbtinsert.c:232
#5  0x0000000000514bb8 in btinsert (rel=0x7f096101a030, values=0x7ffcbd5dbb50, isnull=0x7ffcbd5dbb30, ht_ctid=0x27be708, heapRel=0x7f0961017c00,
    checkUnique=UNIQUE_CHECK_YES, indexInfo=0x27be048) at nbtree.c:205
#6  0x000000000050752a in index_insert (indexRelation=0x7f096101a030, values=0x7ffcbd5dbb50, isnull=0x7ffcbd5dbb30, heap_t_ctid=0x27be708,
    heapRelation=0x7f0961017c00, checkUnique=UNIQUE_CHECK_YES, indexInfo=0x27be048) at indexam.c:212
#7  0x00000000006e5d70 in ExecInsertIndexTuples (slot=0x27be6d8, estate=0x27bd7a8, noDupErr=false, specConflict=0x0, arbiterIndexes=0x0)
    at execIndexing.c:390
#8  0x000000000071f11f in ExecInsert (mtstate=0x27bdb60, slot=0x27be6d8, planSlot=0x27be6d8, estate=0x27bd7a8, canSetTag=true) at nodeModifyTable.c:587
#9  0x0000000000721696 in ExecModifyTable (pstate=0x27bdb60) at nodeModifyTable.c:2175
.......

The following Assert statement in _bt_check_unique fails

  >│386                                     Assert(!_bt_isequal(itupdesc, itup_key, page, offset)); 
   

Upon quick look, it seems like the following git-commit has added above Assert statement:

commit e5adcb789d80ba565ccacb1ed4341a7c29085238
Author: Peter Geoghegan <pg@bowt.ie>
Date:   Wed Mar 20 09:30:57 2019 -0700

    Refactor nbtree insertion scankeys.
   
    Use dedicated struct to represent nbtree insertion scan keys.  Having a
    dedicated struct makes the difference between search type scankeys and
    insertion scankeys a lot clearer, and simplifies the signature of
    several related functions.  This is based on a suggestion by Andrey
    Lepikhov.

....

Including Peter and Hekki in the CC as they are the main author of above git-commit as per the commit message.

-- 

With Regards,

Ashutosh Sharma

EnterpriseDB:http://www.enterprisedb.com