I have got E_RD0043_CACHE_FULL and E_RD0043_CACHE_FULL message
Marek Domachowski
Is some body know how to solve this problem ?
This is a message from Ingres II (AIX 4.3).
The parameter MAX_TBLS was increase from 300 to 1000 but the problem still
exist.
Should I increase it one more time ?
Best regards,
Marek
--
-----------------------------------------------------------------
Lido Technology sp. z o.o.
-----------------------------------------------------------------
Marek Domachowski
e-mail :Marek.Do...@lido-tech.net
tel. : (+48) 58 305 62 00 ext. 26
-----------------------------------------------------------------
Martin Bowes
It sure sounds like the rdf cache is running out of memory, and I would
have expected that expanding the rdf_max_tbls from 300 to 1000 would have
altered the memory allotted (rdf_memory) as it is a derived parameter.
The help available inside CBF would indicate that this is so.
But it doesnt't!!!
So I had a look at the rules file and it appears that the rule for
rdf_memory is based solely on the rdf_blksize. eg:
ii.$.dbms.$.config.rdf_blksiz: 32K;
ii.$.dbms.$.rdf_memory: 340 * ii.$.dbms.$.config.rdf_blksiz, MIN = 4K;
Now, I'm no expert in reading these rules files, but it appears that
without an alteration to the rdf_blksiz parameter, my rdf memory is fairly
stuck on 11141120bytes.
CBF does not allow an alteration to the rdf_blksiz parameter.
So my suggestion is to use CBF to alter the rdf_memory parameter. This is
a derived dbms parameter. I would suggest that you double it and then
mark the parameter as protected.
What have you got to lose?
Martin Bowes
--
Random Duckman Quote #38:
Duckman - Oh Corny, sorry about that little mix up at dinner last night.
Cornfed - The one where you tried to kill me.
Duckman - Lets not quibble over details.
Marek Domachowski
Thank you for your answer :-).
To be precise there are settings of our Ingres installations :
ii.$.dbms.*.rdf_max_tbls: 1000
ii.$dbms.*.rdf_memory: 178257920
ii.$.dbms.*.rdf_tbl_cols: 90
ii.$.dbms.*.rdf_tbl_synonyms: 1
ii.$.dbms.2.rdf_max_tbls: 1000
ii.$.dbms.2.rdf_memory: 44564480
ii.$.dbms.2.rdf_tbl_cols: 90
ii.$.dbms.2.rdf_tbl_synonyms: 1
ii.$.dbms.3.rdf_max_tbls: 1000
ii.$.dbms.3.rdf_memory: 44564480
ii.$.dbms.3.rdf_tbl_cols: 90
ii.$.dbms.3.rdf_tbl_synonyms: 1
ii.$.dbms.private.config.rdf_blksiz: 32768
As you can see all rdf_memory's are set as much more than defined by you
rule.
Questions are :
1) It could be over 11141120bytes ?
2) Can be a different value for each dbms ?
or should I use only one dbms ?
Best regadrs,
Marek
"Martin Bowes" <bo...@bucket.its.unimelb.edu.au> wrote in message
news:2002043023...@bucket.its.unimelb.edu.au...
Martin Bowes
Thats very interesting!
You have three server configurations listed below. Are each of these
servers started? Are they shared cache servers? Do they have a restricted
database list?
Does your RDF error message come up in a particular server and not the
others? Given the size mismatch in rdf_memory between the default server
and the two alternative configurations I'm guessing that you probably see
the error only in a server brought up under the '2' or '3' configuration.
Why do you have rdf_tbl_cols = 90? That seems extraordinarily wide to me!
>
> Hi,
> Thank you for your answer :-).
> To be precise there are settings of our Ingres installations :
> ii.$.dbms.*.rdf_max_tbls: 1000
> ii.$dbms.*.rdf_memory: 178257920
> ii.$.dbms.*.rdf_tbl_cols: 90
> ii.$.dbms.*.rdf_tbl_synonyms: 1
> ii.$.dbms.2.rdf_max_tbls: 1000
> ii.$.dbms.2.rdf_memory: 44564480
> ii.$.dbms.2.rdf_tbl_cols: 90
> ii.$.dbms.2.rdf_tbl_synonyms: 1
> ii.$.dbms.3.rdf_max_tbls: 1000
> ii.$.dbms.3.rdf_memory: 44564480
> ii.$.dbms.3.rdf_tbl_cols: 90
> ii.$.dbms.3.rdf_tbl_synonyms: 1
> ii.$.dbms.private.config.rdf_blksiz: 32768
>
> As you can see all rdf_memory's are set as much more than defined by you
> rule.
> Questions are :
> 1) It could be over 11141120bytes ?
Yes. Like all the memory areas you are at liberty to configure what your
system requires.
> 2) Can be a different value for each dbms ?
Yes. Each dbms server has its own rdf cache. Even if they are shared
servers on the same dmf cache.
However, I dont see much point in making them different sizes if they are
shared servers on the same dmf cache and are not database specific.
> or should I use only one dbms ?
The number of servers isn't affecting the RDF error message.
Martin Bowes
>
> Best regadrs,
> Marek
--
Random Duckman Quote #24:
Cornfed - It was good to have a real client again. Not the pretend kind you
build out of Leggo.
Peter Gale
I'm not sure why you are suffering this problem because, as you say, all
your servers have much more RDF Memory than is the default. And as
Martin points out unless you explicitly changed it there is no way it
gets changed by setting rdf_max_tbls or anything else. Did you
deliberatly change it and if so why?
Is this problem occuring during a particluar query, or does it just
happened after a period of time?
Is the SQL much different from the 6.4 version?
It might be helpful to get some stats out of RDF after you have had this
problem.
Make an sql connection to the DBMS server that is giving the problem
(II_DBMS_SERVER=...)
Specify a trace output file
SET TRACE OUTPUT 'filename'\g
And then shut down ingres. In the output file you will find a section on
RDF Statistics and in particular this bit.
! GENERAL STATSISTICS
!
!
+-----------------------------------------------------------------------
----+
! : RESOURCE REQUESTS : ERROR CONDITIONS
:
!
+-------------------------+---------------------------------------------
----:
! :Avg Col/Table 3 : Exception Cnt 0: No Memory
0:
! :Avg Idx/Table 0 : Bad Unfix Req 0: Unfix Errors
0:
! :Tables Cached 4 :
:
This might give some clues as to how much of a load you are putting on
rdf and whether your parameter settings are correct (e.g. rdf_tbl_cols =
90)
I would go with Martins suggestion of doubling the memory. But I would
also suggest that you don't need 3 DBMS server definitions. As you are
running on AIX you can use OS Threads (II_THREAD_TYPE=EXTERNAL) and turn
off slaves (make sure II_NUM_SLAVES is not set). This is not going to
help directly with the RDF problem but it will reduce the overall
demands on memory.
HTH
Peter Gale
PJG Computer Services Ltd
peter...@pjg.uk.com
Richard Yates
> [ ... getting ... ] RDF Statistics and in particular this bit.
>
> ! GENERAL STATSISTICS
>
+-----------------------------------------------------------------------
> ----+
> ! : RESOURCE REQUESTS : ERROR CONDITIONS
> ----:
> ! :Avg Col/Table 3 : Exception Cnt 0: No Memory
> 0:
> ! :Avg Idx/Table 0 : Bad Unfix Req 0: Unfix Errors
> 0:
> ! :Tables Cached 4 :
> :
Is that the same stuff as you get at the end of the iidbms log
on shutdown (using ingsetenv II_DBMS_LOG iidbms_%p.log or similar) ?
There's a whole rake of information on RDF and other things (for
general interest I'm appending RDF stats from an II_DBMS_LOG below).
I've mentioned several times (and raised an issue with CA) that
it would be very nice to have explanations of the stats. produced
in this log file, to no avail. Perhaps it's secret (or only a
couple of people are party to this knowledge)!
Richard.
!-------------------------- RDF Statistics Report:
----------------------------
-
! CACHE ALLOCATION/TUNING INFORMATON
!
! :---------------------------- HARD LIMITS
--------------------------------:
! :cache size in bytes : 11141120 [Tune w/ RDF_MEMORY
parameter] :
! :max tables on cache : 4096 [Tune w/ RDF_MAX_TABLES
parameter] :
! :max QRY Tree obects : 4096 [Tune w/ RDF_MAX_TABLES
parameter] :
! :max LDB descriptors : 0 [Tune w/ RDF_MAX_COL_DEFAULTS
parameter]:
! :max defaults on cache: 19 [Tune
as :
! : (RDF_MAX_TABLES *
RDF_TABLE_COLUMNS)/5 ]:
! :------------------------- STARTUP
PARAMETERS-------------------------------:
! :rdf_memory : 11141120 rdf_table_synonyms:
1 :
! :rdf_max_tables : 4096 rdf_cache_ddbs :
0 :
! :rdf_max_col_default: 19 rdf_average_ldbs :
0 :
!
:---------------------------------------------------------------------------:
! CATALOG OPEATION STATISTICS
!
!
+-------------+-------------------------------------------------------------+
! : OPERATION : REQUEST ON CACHE BUILT NOT FOUND CAT APPEND CAT
DELETE:
! +-------------:--------- --------- --------- --------- ----------
----------+
! :Existance Chk: 8392 3195 2434 2763
N/A N/A:
! :Core Catalog : 7379487 6649976 153425 576359
N/A N/A:
! :Synonym : 579420 N/A 333 578959
317 318:
! :Statistics : 9112523 9094449 18679 -605
N/A N/A:
! :Integrity : 0 0 0 0
0 0:
! :View : 103861 100624 3235 2
0 0:
! :Procedure : 1385 4 1363 18
4 8:
! :Rule : 288570 286997 1573 0
4 4:
! :Event Permits: 680 N/A 680 0
0 0:
! :Permit Tree : 0 0 0 0
2483 14:
! :Security Alrm: 0 0 0 0
0 0:
! :All To All : N/A N/A N/A N/A
0 0:
! :Retr to All : N/A N/A N/A N/A
0 0:
! :Instal Cats : N/A N/A N/A N/A
18 0:
! :Comments Cat : N/A N/A N/A N/A
0 0:
!
! QRYMOD OPEATION STATISTICS
!
!
+-----------------+---------------------------------------------------------+
! : :
MULTIPLE:
! : OPERATION : REQUEST ON CACHE BUILT NOT FOUND READ
REQ'D:
! +-----------------: ---------- ---------- ---------- ----------
----------:
! :Protection Tuples: 530815 507488 1072 0
22254:
! :Integrity Tuples : 55 0 0
55 0:
! :Rule Tuples : 121296 120808 488
0 0:
! :Proc Param Tups : 0 0 0
0 0:
! :Key Tuples : 0 0 0
0 0:
! :Event Permit Tups: 0 N/A 0
0 0:
! :Alarm Tuples : 0 0 0
0 0:
!
! GENERAL STATSISTICS
!
!
+---------------------------------------------------------------------------+
! : RESOURCE REQUESTS : ERROR
CONDITIONS :
!
+-------------------------+-------------------------------------------------:
! :Avg Col/Table 21 : Exception Cnt 0: No
Memory 0:
! :Avg Idx/Table 0 : Bad Unfix Req 0: Unfix
Errors 0:
! :Tables Cached 155862
: :
!
! MEMORY MANAGEMENT STATISTICS
!
!
+------------------------------+-------------------------------------------+
! : FIX : UNFIX SHARED
PRIVATE :
! +------------------------------+ ---------------- ----------
-----------:
! :Rel Cache Object 6815520 : Rel Cache Object
6743696 60:
! :Integrity Tree Obj 0 : Integrity Tree Obj
0 0:
! :Procedure Object 2481 : Procedure Object
1367 0:
! :Permit Tree Object 0 : Permit Tree Object
0 0:
! :Rule Tree Object 288570 : Rule Tree Object
288570 0:
!
!
+--------------------------------------------------------------------------+
! : LOCK FOR
UPDATE :
!
+--------------------------------------------------------------------------+
! : Shared Cache Object 2279442 : Shared Cache Overflow
Ctr 0 :
! : Private Cache Object 101 : Private Cache Overflow
Ctr 0 :
!
!
+--------------------------------------------------------------------------+
! : CACHE INVALIDATION
REQUESTS :
!
+------------------------------------+-------------------------------------+
! : Single Object 93274 : Tree
Class 0 :
! : Procedure Object 41 : Tree
Object 2 :
! : RELATION Cache 0 : LDBDESC Cache
0 :
! : QTREE Cache 0 : Obj not on cache
996 :
!
!
+------------------------------------+-------------------------------------+
! : MEMORY RECLAIM : MEMORY
REFRESH :
!
+------------------------------------+-------------------------------------+
! : Rel Cache 27 : Rel cache Object
15016 :
! : QT Cache 1165 : QT Cache Object
1114 :
! : LDB Cache 0
: :
!
! : Defaults Cache 0
: :
!
!note: a single table may show up several times in the Tables Cached
count if it
is
! flushed from cache several times.
!
!
!---------------------------- End RDF Statistics
----------------------------
--
Erratic news-swerver! :. post+e-mail to be sure I get your msg.
The Open University is not responsible for content herein, which
may be incorrect and is used at reader's own risk.
Peter Gale
if you forgot to set II_DBMS_LOG.
The extract was indeed a very small extract from a very large amount of
information.
Peter Gale
PJG Computer Services Ltd
peter...@pjg.uk.com
>-----Original Message-----
>From: owner-in...@ams.org
>[mailto:owner-in...@ams.org] On Behalf Of Richard Yates
>Sent: 01 May 2002 11:36
>To: info-...@ams.org
>Subject: Re: I have got E_RD0043_CACHE_FULL and
>E_RD0043_CACHE_FULL message
>
>
>!
>+-------------------------+------------------------------------
>---------
>+-------------------------+----:
Marek Domachowski
A oto co napisal Peter Gale. Mysle ze ta sprawa moznaby sie zajac w ramach
kursu :-)
Marek
----- Original Message -----
From: "Peter Gale" <peter...@pjg.uk.com>
To: "'Marek Domachowski'" <marek.do...@lido-tech.net>
Cc: "'ingres newsgroup'" <info-...@ams.org>; "'Martin Bowes'"
<bo...@bucket.its.unimelb.edu.au>
Sent: Wednesday, May 01, 2002 9:47 AM
Subject: RE: I have got E_RD0043_CACHE_FULL and E_RD0043_CACHE_FULL message
> Marek,