ORA-12853 and ORA-04031 simultaneously

Question

Hello,
I am getting the following errors when I try to start Oracle
ORA-12853: insufficient memory for PX buffers: current 0K, max needed 2640K
ORA-04031: unable to allocate 21544 bytes of shared memory ("large pool","unknown object","large pool","PX msg pool")
My init.ora file is as follows -
EDA.__db_cache_size=1174405120
EDA.__java_pool_size=33554432
EDA.__large_pool_size=0
EDA.__shared_pool_size=1174405120
EDA.__streams_pool_size=0
*._OPTIM_PEEK_USER_BINDS=FALSE
*.background_dump_dest='/oracle/EDA/saptrace/background'
*.compatible='10.2.0'
*.control_file_record_keep_time=30
*.control_files='/oracle/EDA/origlogA/cntrl/cntlrEDA.dbf','/oracle/EDA/origlogB/cntrl/cntrlEDA.dbf','/oracle/EDA/sapdata1/cntrl/cntrlEDA.dbf'
*.core_dump_dest='/oracle/EDA/saptrace/background'
*.db_block_size=8192
*.db_cache_size=1159641169
*.db_files=254
*.db_name='EDA'
*.dml_locks=4000
*.event='10191 trace name context forever, level 1'
*.FILESYSTEMIO_OPTIONS='setall'
*.job_queue_processes=1
*.log_archive_dest='/oracle/EDA/oraarch/EDAarch'
*.log_buffer=1048576
*.log_checkpoint_interval=0
*.log_checkpoints_to_alert=true
*.open_cursors=800
*.optimizer_features_enable='10.2.0.1'
*.pga_aggregate_target=1546188226
*.processes=80
*.recyclebin='off'
*.remote_login_passwordfile='exclusive'
*.remote_os_authent=true
*.sessions=96
*.sga_max_size=2319282339
*.sga_target=2399141888
*.shared_pool_reserved_size=115964116
*.shared_pool_size=1159641169
*.sort_area_retained_size=0
*.sort_area_size=2097152
*.statistics_level='typical'
*.undo_management='AUTO'
*.undo_retention=43200
*.undo_tablespace='PSAPUNDO'
*.user_dump_dest='/oracle/EDA/saptrace/usertrace'
*.workarea_size_policy='AUTO'
According to me everything looks fine. Kindly help me sort this error.
Thanks in advance.

Answer 1

1. Play with Shared Pool....
a. The following query determines the available memory for SHARED_POOL_SIZE in Oracle sga
select sum (bytes)/1024/1024 from v$sgastat where pool=u2019shared poolu2019
b. The following query determines the total used memory by shared_Pool in Oracle SGA.
select sum (bytes)/1024/1024 from v$sgastat where pool=u2019shared poolu2019 and name not in (u2019free memoryu2019)
c. This is the most important query
select
sum(a.bytes)/(1024*1024))shared_pool_used,
max(b.value)/(1024*1024) shared_pool_size,
sum(a.bytes)/(1024*1024))-
(sum(a.bytes)/(1024*1024)) shared_pool_avail,
((sum(a.bytes)/(10241024))/(max(b.value)/(10241024)))*100
pct_shared_pool_avl
from v$sgastat a, v$parameter b
where (a.pool=u2019shared poolu2019
and a.name not in (u2019free memoryu2019))
and
b.name=u2019shared_pool_sizeu2019
You need to continously monitor the shared Pool with the above query at differnet times. During Peak times and Non peak times to have glance of shared pool usage in the Oracle database.
if the available pct_shared_pool_avl crosses 95% then i think you should re-consider the Process of increasing the shared_pool_size.
2. There are many way to improve Shared Pool performance.
a. Ask ABAPers to write more generic and reusablecode.
b. Using of right block size.
c. Proper design of the database.
Comment on ORA-04031 :
This error should not appear in any of the application logs, the alert log or any trace files. Do not depend on ORA-04031 errors being written to the alert log, as 4031 errors only appear in the alert log if they affect background process operations (such as PMON activities). 4031u2019s are not internal errors and so could be trapped and handled by the application (this is not recommended).
From 10gR1 onwards, a 4031 trace file is written to the user_dump_dest (or background_dump_dest) directory; this trace file is useful in diagnosing the nature of problem
Hope this will help you.
Regards,
SK
OCP DBA -9i,10g