Re: Microvacuum support for Hash Index

Hi Ashutosh,

On 01/06/2017 12:54 AM, Ashutosh Sharma wrote:
>> using pgbench -M prepared -c 10 -j 10 -T 600 -f test.sql test
>>
>> crashes after a few minutes with
>>
>> TRAP: FailedAssertion("!(LWLockHeldByMeInMode(((LWLock*)
>> (&(bufHdr)->content_lock)), LW_EXCLUSIVE))", File: "bufmgr.c", Line: 3781)
>
> Attached v4 patch fixes this assertion failure.
>

Yes, that fixes the problem.

However (master / WAL v7 / MV v4),

--- ddl.sql ---
CREATE TABLE test AS SELECT generate_series(1, 10) AS id, 0 AS val;
CREATE INDEX IF NOT EXISTS idx_id ON test USING hash (id);
CREATE INDEX IF NOT EXISTS idx_val ON test USING hash (val);
ANALYZE;
--- ddl.sql ---

--- test.sql ---
\set id random(1,10)
\set val random(0,10)
BEGIN;
DELETE FROM test WHERE id = :id;
INSERT INTO test VALUES (:id, :val);
COMMIT;
--- test.sql ---

gives

#9  0x000000000098a83e in elog_finish (elevel=20, fmt=0xb6ea92 
"incorrect local pin count: %d") at elog.c:1378
#10 0x00000000007f0b33 in LockBufferForCleanup (buffer=1677) at 
bufmgr.c:3605
#11 0x0000000000549390 in XLogReadBufferForRedoExtended 
(record=0x2afced8, block_id=1 '\001', mode=RBM_NORMAL, 
get_cleanup_lock=1 '\001', buf=0x7ffe3ee27c8c) at xlogutils.c:394
#12 0x00000000004c5026 in hash_xlog_vacuum_one_page (record=0x2afced8) 
at hash_xlog.c:1109
#13 0x00000000004c5547 in hash_redo (record=0x2afced8) at hash_xlog.c:1214
#14 0x000000000053a361 in StartupXLOG () at xlog.c:6975
#15 0x00000000007a4ca0 in StartupProcessMain () at startup.c:216

on the slave instance in a master-slave setup.

Also, the src/backend/access/README file should be updated with a 
description of the changes which this patch provides.

Best regards,
  Jesper