Doubt in heap_release_fetch

Doubt in heap_release_fetch

Hello,

What do the following lines mean :

  /* Tuple failed time qual, but maybe caller wants to see it anyway. */
        if (keep_buf)
                *userbuf = buffer;
        else
        {
                ReleaseBuffer(buffer);
                *userbuf = InvalidBuffer;
        }

What is the time qualification check ?

Thanks,
Suresh


Never miss a thing.  Make Yahoo your homepage.

Checking stack depth

Hello,

Why is check_stack_depth function enforced in context of evaluating expressions in PostgreSQL ? What sort of recursion we are trying to safeguard ?

thanks,
Suresh

Re: Doubt in IndexScanDescData

Re: Checking stack depth

Suresh  writes:
> Why is check_stack_depth function enforced in context of evaluating expressions in PostgreSQL ? What sort of recursion we are trying to safeguard ?

create function foo(int) returns int as $$
select foo($1) $$ language sql;

select foo(1);
		regards, tom lane

Doubt in index subplan query

Hello,

I have a query plan for a certain query

 Nested Loop  (cost=100000000.00..38761761090.50 rows=3000608 width=8)
   ->  Seq Scan on lineitem  (cost=100000000.00..100213649.15 rows=6001215 width=8)
   ->  Index Scan using oindex2 on myorders  (cost=0.00..6442.27 rows=1 width=4)
         Index Cond: ("outer".l_orderkey = myorders.o_orderkey)
         Filter: (subplan)
         SubPlan
           ->  Index Scan using cnation on customer  (cost=0.00..12859.39 rows=5251 width=0)
                 Index Cond: (c_nationkey = 10)

How is the subplan handled by postgres at index level ? Is any sort of hashing done ?

Thanks and regards,
Suresh

Re: Stack depth exceeded error

Hi,

The code uses Asynchronous I/O for fetching certain tids. The code works fine if I use only one condition in where condition, but fails if I use multiple condition.

--
Suresh Iyengar

Gregory Stark <stark@enterprisedb.com> wrote:
 "Suresh"  writes:

> Hello,
>
> I have custom postgres code. 

What kind of code is this? The error below is typical if you create new
threads in the server.


-- 
 Gregory Stark
 EnterpriseDB http://www.enterprisedb.com
 Ask me about EnterpriseDB's 24x7 Postgres support!

Be a better friend, newshound, and know-it-all with Yahoo! Mobile.  Try it now.

Doubt in IndexScanDescData

[ "include/access/relscan.h" ]

In  IndexScanDescData,  whats the purpose of having two Relation variables.

typedef struct IndexScanDescData
{
        Relation        heapRelation;   /* heap relation descriptor, or NULL */
        Relation        indexRelation;  /* index relation descriptor */
....
...
}IndexScanDescData;


Be a better friend, newshound, and know-it-all with Yahoo! Mobile.  Try it now.

Getting statistics

Hello,

I want to collect various statistics like time taken, number of context switches, page faults etc.. for a query being run. postgres.c contains lots of getrusage related things.

Is there any way to calculate all the things without writing any custom code ?

Thanks and regards,
Suresh

Be a better friend, newshound, and know-it-all with Yahoo! Mobile.  Try it now.

Doubt in index scan code

Hello all,

I have a custom code written inside postgres in an application we use.
The snippet is as below :
Here inner plan is an index scan.

  scandesc = ((IndexScanState *)innerPlan)->iss_ScanDesc; 

  flag=index_getmulti(scandesc, &tidelm->tid, 1, &ret_tids);

Now consider a query like

explain  select * from dept,manager where did=id ;
                                QUERY PLAN                                 
---------------------------------------------------------------------------
 Nested Loop  (cost=0.00..269.09 rows=45 width=72)
   -> seq scan on manager  (cost=0.00..6.50 rows=45 width=36)
   ->  Index Scan using id1 on dept  (cost=0.00..5.82 rows=1 width=36)
         Index Cond: ("outer".did = dept.id)

Say seq scan retrieves did in the order 30,10, 20.. My doubt is in what order 
will index_getmulti return tids. How does the scandesc work ? 

Will it return the tids as firstly macthing inners for dept=30, then dept=10 ?

Please help me with this.

Thanks and regards,
Suresh










Be a better friend, newshound, and know-it-all with Yahoo! Mobile.  Try it now.

segfault in locking code

Hello,

I have a custom code in postgres which runs properly in some occasions and segfaults some times. The trace is as below :

Program received signal SIGSEGV, Segmentation fault.
0x081ae8c4 in LWLockRelease (lockid=664)
    at ../../../../src/include/storage/s_lock.h:128
128             __asm__ __volatile__(


(gdb) where
#0  0x081ae8c4 in LWLockRelease (lockid=664)
    at ../../../../src/include/storage/s_lock.h:128
#1  0x0808f820 in heap_fetch_tuple (relation=0xb5d986d8, snapshot=0xa298aa0, 
    buffer=305, tid=0xa23f600, tuple=0xa29db0c, pgstat_info=0xa29db30, 
    tupStat=0xbfac9374) at heapam.c:3404
#2  0x08144df2 in ExecNestLoop (node=0xa298f30) at nodeNestloop.c:452
#3  0x08136840 in ExecProcNode (node=0xa298f30) at execProcnode.c:352
#4  0x08135ba1 in ExecutorRun (queryDesc=0xa298ac8, 
    direction=ForwardScanDirection, count=0) at execMain.c:1162
#5  0x081b7e60 in PortalRunSelect (portal=0xa296a98, 
    forward=<value optimized out>, count=0, dest=0x82d3308) at pquery.c:794
#6  0x081b8a88 in PortalRun (portal=0xa296a98, count=2147483647, 
    dest=0x82d3308, altdest=0x82d3308, completionTag=0xbfac9608 "")
    at pquery.c:646
#7  0x081b48fc in exec_simple_query (
    query_string=0xa275b58 "select l_orderkey as a from tpcd.orders, tpcd.lineitem where o_orderkey=l_orderkey ;\n") at postgres.c:1003
#8  0x081b6371 in PostgresMain (argc=1, argv=0xa2379f0, 
    username=0xa238398 "suresh") at postgres.c:3221
#9  0x081532e3 in main (argc=2, argv=Cannot access memory at address 0xfffffffd
) at main.c:411

It segfaults in the locking _asm_ code. What could be the reason for this variable behavior ?

Thanks and regards,
Suresh
Never miss a thing.  Make Yahoo your homepage.

Re: Doubt in heap_release_fetch

Suresh  writes:
> What is the time qualification check ?

HeapTupleSatisfiesVisibility().  See
src/include/utils/tqual.h
src/backend/utils/time/tqual.c
and if none of this is making any sense maybe you need to start here:
http://developer.postgresql.org/pgdocs/postgres/mvcc.html
		regards, tom lane

Re: Doubt in index subplan query

On Jun 20, 2008, at 1:11 AM, Suresh wrote:
> I have a query plan for a certain query
>
>  Nested Loop  (cost=100000000.00..38761761090.50 rows=3000608 width=8)
>    ->  Seq Scan on lineitem  (cost=100000000.00..100213649.15  
> rows=6001215 width=8)
>    ->  Index Scan using oindex2 on myorders  (cost=0.00..6442.27  
> rows=1 width=4)
>          Index Cond: ("outer".l_orderkey = myorders.o_orderkey)
>          Filter: (subplan)
>          SubPlan
>            ->  Index Scan using cnation on customer   
> (cost=0.00..12859.39 rows=5251 width=0)
>                  Index Cond: (c_nationkey = 10)
>
> How is the subplan handled by postgres at index level ? Is any sort  
> of hashing done ?

This is better asked on pgsql-general... but the subplan does exactly  
what it says; an index scan. It will be executed for every row of the  
calling query.
-- 
Decibel!, aka Jim C. Nasby, Database Architect  decibel@decibel.org
Give your computer some brain candy! www.distributed.net Team #1828

Stack depth exceeded error

Hello,

I have custom postgres code. I get the error below for the query 

"select l_orderkey as a from tpcd.orders, tpcd.lineitem where o_orderkey=l_orderkey and l_partkey<100 and l_linestatus='F';"

ERROR:  stack depth limit exceeded
HINT:  Increase the configuration parameter "max_stack_depth".

However, the same code runs fine with one condition in where clause, but fails with the error above in case of multiple conditions.

Whats the cause of this error ? I tried increasing the stack limit; but it doesnt help.

--
Suresh Iyengar
Be a better friend, newshound, and know-it-all with Yahoo! Mobile.  Try it now.

Re: Doubt in IndexScanDescData

Re: Stack depth exceeded error

"Suresh"  writes:

> Hello,
>
> I have custom postgres code. 

What kind of code is this? The error below is typical if you create new
threads in the server.


--  Gregory Stark EnterpriseDB          http://www.enterprisedb.com Ask me about EnterpriseDB's 24x7 Postgres support!

Re: Doubt in IndexScanDescData

One thing you might be missing is that indexes are relations too. They're a
bit different than heap relations (ie "tables") but they share enough
structure that it's worth using the same datatypes to represent both.

--  Gregory Stark EnterpriseDB          http://www.enterprisedb.com Ask me about EnterpriseDB's RemoteDBA services!

Re: segfault in locking code

Suresh  writes:
> I have a custom code in postgres which runs properly in some occasions and segfaults some times. The trace is as below :

The traceback you show appears to lead through code that doesn't exist
in any public version of Postgres.  So I think it's your own bug to
solve.
		regards, tom lane

Re: Doubt in index scan code

Re: Doubt in IndexScanDescData