PLSQL function does not return the correct number of rows?

Question

Hey folks. I'm still green when it comes to writing PLSQL. It's fun, rewarding and very frustrating. Hence, I'm turning to the experts. If you folks can help me understand what I'm doing wrong here, I'd really appreciate it.
The code is somewhat specific to my company's product, but I think it should be easy to read and understand what I'm doing. If not, please let me know what I can clarify.
All i'm trying to do is determine if the most recent iteration of data available for a particular host is a full scan or not (level2). I go about this in the following manner:
1. get the operatingsystem id, it's scandate (preferred), the most recent scandate and it's scan status from a table of where all operating systems data lives. Loop through all the Oses
(from this I set v_osid, v_mostrecentscandate, v_scandate).
2. Before doing the crazy logic, pick the low hanging fruit
2a. if the the level2 status of the host is N, then v_level2 = 'N';
2b. if the level2 = 'Y' and the mostrecentscandate and scandate are identical, then v_level2 = 'Y';
2c. for all other cases, go to 3
3. Using v_mostrecentscandate, find all table id that may hold the most recent instance of data for the host
4. Loop through through the concatenation of that id + _base. If you find the id in those tables, then store the id for the next step.
5. When you I find the right id, I now concatenate the id + attrdata. For the host id, I look for any rows where attribute_value in (..) and the corresponding number_value is not null.
5b. set v_level2 = 'Y'
5c. otherwise, set v_level2 = 'N'
6 end the loop
7 wash, rinse, repeat for each OS.
create or replace package body mostrecentlevel2 as
function getMostRecentL2 return bdna_mostrecent_level2 pipelined IS
v_lsid NUMBER;
v_sql VARCHAR2(5000);
v_sql_baseid NUMBER;
v_sql_numv NUMBER;
v_lsidt VARCHAR2(5000);
v_lsidt2 VARCHAR2(5000);
v_sql_rec VARCHAR2(5000);
v_osid NUMBER;
v_anchor DATE;
v_ls CHAR(2);
v_level2 CHAR(1);
v_mostrecentscandate DATE;
v_scandate DATE;
cursor getOSinfo_cur is select operatingsystem_id, scandate, mostrecentscandate, level2 from bdna_all_os;
cursor getlsID_cur is select id from local_scan where
          ((trunc(collect_start_time) - to_date(v_anchor))*24*60*60) <= ((to_date(v_mostrecentscandate) - to_date(v_anchor))*24*60*60)
          and
          ((trunc(collect_end_time) - to_date(v_anchor))*24*60*60) >= ((to_date(v_mostrecentscandate) - to_date(v_anchor))*24*60*60);
getOSinfo_rec getOSinfo_cur%rowtype;
getlsID_rec getlsID_cur%rowtype;
BEGIN
v_ls := 'ls';
v_anchor := '01-JAN-01';
FOR getOSinfo_rec IN getOSinfo_cur LOOP
     v_osid := getOSinfo_rec.operatingsystem_id;
     v_mostrecentscandate := getOSinfo_rec.mostrecentscandate;
     v_scandate := getOSinfo_rec.scandate;
     IF getOSinfo_rec.level2 = 'N' THEN
          v_level2 := 'N';
     ELSIF getOSinfo_rec.level2 = 'Y' THEN
          IF v_mostrecentscandate != v_scandate THEN
               FOR getlsID_rec IN getlsID_cur LOOP
                    v_lsid := getlsID_rec.id;
                    v_lsidt := v_ls||v_lsid;
                    v_sql := 'select id from '||v_lsidt||'_base where id = '||chr(39)||v_osid||chr(39);
                    EXECUTE IMMEDIATE v_sql into v_sql_baseid;
                    IF SQL%ROWCOUNT > 0 THEN
                         v_lsidt2 := v_lsidt;
                         v_sql := '';
                    END IF;
               END LOOP;
               v_sql := 'select number_value from '||v_lsidt2||'_attr_data where
                         lower(attribute_name) IN ('||chr(39)||'numcpus'||chr(39)||', '||chr(39)||'totalmemory'||chr(39)||', '||chr(39)||'cpuutilpercent'||chr(39)||', '||chr(39)||'numprocesses'||chr(39)||')
                         and
                         number_value is not NULL
                         and
                         element_id = '||chr(39)||v_osid||chr(39);
               EXECUTE IMMEDIATE v_sql into v_sql_numv;
               IF SQL%ROWCOUNT > 0 THEN
                    v_level2 := 'Y';
               ELSE v_level2 := 'N';
               END IF;
          END IF;
          v_level2 := 'Y';
     END IF;
     PIPE ROW (mostRecentLevel2Format(v_osid,v_mostrecentscandate,v_level2));
END LOOP;
END;
END;
/Now some will ask why I'm using pipelining? Again, I'm green.. I was reading around, looking for a way to make this code run as fast as possible (because it's potentially got to go through 56K records and perform the expensive work on).
I also realize I'm not providing the type or package code, and that's because I think I'm good on that. The code above compiles just fine without errors and when it runs, it only returns 6 consecutive rows.. I'm expecting 70K lol. So I know I'm doing something wrong.
Any thoughts?
Oh forgot to add this is on 11g R1 Enterprise Edition
Edited by: ErrolDC on Nov 14, 2011 4:52 PM
Edited by: ErrolDC on Nov 14, 2011 5:07 PM

Answer 1

ErrolDC wrote:
Hey folks. I'm still green when it comes to writing PLSQL. It's fun, rewarding and very frustrating. Hence, I'm turning to the experts. If you folks can help me understand what I'm doing wrong here, I'd really appreciate it.
The code is somewhat specific to my company's product, but I think it should be easy to read and understand what I'm doing. If not, please let me know what I can clarify.Post a complete script that peoople who aren't as familiar with the application as you are can run to re-create the problem and test their ideas. In this case, that includes CREATE TABLE and INSERT statements for the tables used (just the columns needed for this job), a query that uses the function, and the results you want from that query given the data you posted.
All i'm trying to do is determine if the most recent iteration of data available for a particular host is a full scan or not (level2). I go about this in the following manner:
1. get the operatingsystem id, it's scandate (preferred), the most recent scandate and it's scan status from a table of where all operating systems data lives. Loop through all the Oses
(from this I set v_osid, v_mostrecentscandate, v_scandate).
2. Before doing the crazy logic, pick the low hanging fruit
2a. if the the level2 status of the host is N, then v_level2 = 'N';
2b. if the level2 = 'Y' and the mostrecentscandate and scandate are identical, then v_level2 = 'Y';
2c. for all other cases, go to 3
3. Using v_mostrecentscandate, find all table id that may hold the most recent instance of data for the host
4. Loop through through the concatenation of that id + _base. If you find the id in those tables, then store the id for the next step.
5. When you I find the right id, I now concatenate the id + attrdata. For the host id, I look for any rows where attribute_value in (..) and the corresponding number_value is not null.
5b. set v_level2 = 'Y'
5c. otherwise, set v_level2 = 'N'
6 end the loop
7 wash, rinse, repeat for each OS.
create or replace package body mostrecentlevel2 as
function getMostRecentL2 return bdna_mostrecent_level2 pipelined IS
v_lsid NUMBER;
v_sql VARCHAR2(5000);
v_sql_baseid NUMBER;
v_sql_numv NUMBER;
v_lsidt VARCHAR2(5000);
v_lsidt2 VARCHAR2(5000);
v_sql_rec VARCHAR2(5000);
v_osid NUMBER;
v_anchor DATE;
v_ls CHAR(2);
v_level2 CHAR(1);
v_mostrecentscandate DATE;
v_scandate DATE;
cursor getOSinfo_cur is select operatingsystem_id, scandate, mostrecentscandate, level2 from bdna_all_os;
cursor getlsID_cur is select id from local_scan where
          ((trunc(collect_start_time) - to_date(v_anchor))*24*60*60) <= ((to_date(v_mostrecentscandate) - to_date(v_anchor))*24*60*60)
          and
          ((trunc(collect_end_time) - to_date(v_anchor))*24*60*60) >= ((to_date(v_mostrecentscandate) - to_date(v_anchor))*24*60*60);
getOSinfo_rec getOSinfo_cur%rowtype;
getlsID_rec getlsID_cur%rowtype;
BEGIN
v_ls := 'ls';
v_anchor := '01-JAN-01';
FOR getOSinfo_rec IN getOSinfo_cur LOOP
     v_osid := getOSinfo_rec.operatingsystem_id;
     v_mostrecentscandate := getOSinfo_rec.mostrecentscandate;
     v_scandate := getOSinfo_rec.scandate;
     IF getOSinfo_rec.level2 = 'N' THEN
          v_level2 := 'N';
     ELSIF getOSinfo_rec.level2 = 'Y' THEN
          IF v_mostrecentscandate != v_scandate THEN
               FOR getlsID_rec IN getlsID_cur LOOP
                    v_lsid := getlsID_rec.id;
                    v_lsidt := v_ls||v_lsid;
                    v_sql := 'select id from '||v_lsidt||'_base where id = '||chr(39)||v_osid||chr(39);
                    EXECUTE IMMEDIATE v_sql into v_sql_baseid;
                    IF SQL%ROWCOUNT > 0 THEN
                         v_lsidt2 := v_lsidt;
                         v_sql := '';
                    END IF;
               END LOOP;
               v_sql := 'select number_value from '||v_lsidt2||'_attr_data where
                         lower(attribute_name) IN ('||chr(39)||'numcpus'||chr(39)||', '||chr(39)||'totalmemory'||chr(39)||', '||chr(39)||'cpuutilpercent'||chr(39)||', '||chr(39)||'numprocesses'||chr(39)||')
                         and
                         number_value is not NULL
                         and
                         element_id = '||chr(39)||v_osid||chr(39);
               EXECUTE IMMEDIATE v_sql into v_sql_numv;
               IF SQL%ROWCOUNT > 0 THEN
                    v_level2 := 'Y';
               ELSE v_level2 := 'N';
               END IF;
          END IF;
          v_level2 := 'Y';
     END IF;
     PIPE ROW (mostRecentLevel2Format(v_osid,v_mostrecentscandate,v_level2));
END LOOP;
END;
END;
/Now some will ask why I'm using pipelining? Again, I'm green.. I was reading around, looking for a way to make this code run as fast as possible (because it's potentially got to go through 56K records and perform the expensive work on).
I also realize I'm not providing the type or package code, and that's because I think I'm good on that. The code above compiles just fine without errors and when it runs, it only returns 6 consecutive rows.. I'm expecting 70K lol. So I know I'm doing something wrong. You're calling TO_DATE with a DATE argument. Why are you calling TO_DATE at all?
It seems like this condition:
((trunc(collect_start_time) - to_date(v_anchor))*24*60*60) <= ((to_date(v_mostrecentscandate) - to_date(v_anchor))*24*60*60) is equivalent to
TRUNC (collect_start_time) <= v_mostrecentscandateThat probably has nothing to do with why you're only getting 6 rows.