Stored procedure hangs seemingly without explanation - tsql

we have a stored procedure that ran fine until 10 minutes ago and then it just hangs after you call it.
Observations:
Copying the code into a query window yields the query result in 1 second
SP takes > 2.5 minutes until I cancel it
Activity Monitor shows it's not being blocked by anything, it's just doing a SELECT.
Running sp_recompile on the SP doesn't help
Dropping and recreating the SP doesn't help
Setting LOCK_TIMEOUT to 1 second does not help
What else can be going on?
UPDATE: I'm guessing it had to do with parameter sniffing. I used Adam Machanic's routine to find out which subquery was hanging. I found things wrong with the query plan thanks to the hint by Martin Smith. I learned about EXEC ... WITH RECOMPILE, OPTION(RECOMPILE) for subqueries within the SP, and OPTION (OPTIMIZE FOR (#parameter = 1)) in order to attack parameter sniffing. I still don't know what was wrong in this particular case but I came out of this battle seasoned and much better armed. I know what to do next time. So here's the points!

I think that this is related to parameter sniffing and the need to parameterize your input params to local params within the SP. Adding with recompile causes the execution plan to be recreated and eliminates much of the benefits of having a SP. We were using With Recompile on many reports in an attempt to eliminate this hanging issue and it occassionally resulted in hanging SP's that may have been related to other locks and/or transactions accessing the same tables simultaneously. See this link for more details
Parameter Sniffing (or Spoofing) in SQL Server and change your SP's to the following to fix this:
CREATE PROCEDURE [dbo].[SPNAME] #p1 int, #p2 int
AS
DECLARE #localp1 int, #localp2 int
SET #localp1=#p1
SET #localp2=#p2

Run Adam Machanic's excellent sp_WhoIsActive stored proc while your query is running. It'll give you the wait information - meaning, what the stored proc is waiting on - plus things like the execution plan:
http://www.brentozar.com/archive/2010/09/sql-server-dba-scripts-how-to-find-slow-sql-server-queries/
If you want the outer command (like a calling stored procedure's full text), use the #get_outer_command = 1 parameter as well.

First thing First.
Please check if there are any uncommitted transactions. A begin transaction without "COMMIT TRANSACTION"

Thanks for all comments.
I still haven't found the answer, but I will post the progress here.
I failed to reproduce the problem before, but today I chanced upon another stored procedure with the same problem. Again the same symptoms appeared:
Hanging piece of query runs fine and quick (3 secs) in normal query window (hanging piece identified with sp_whoisactive)
No locks, according to Activity Monitor SPID is doing SELECT
Stored procedure runs for over 6 hours without response
Parameters passed to SP and variables declared in window are the same
Using above hints, I found the SP execution plan and it showed nothing out of the ordinary (to me, at least). Creating a new stored procedure with same contents did not solve the problem either. So I started stripping the SP to less and less contents until I encountered a UDF call to another database. When I removed that (replaced the call by the inline contents of the function, a CASE statement), it ran fine again.
So this COULD have been the problem, but I am not very certain, as last time the problem disappeared by itself and I also changed a lot of other things while stripping this SP.

When we add new data sometimes the execution plan becomes invalid or out of date then the stored procedure starts going into this limbo phase. Run the following commands on your database
DBCC DROPCLEANBUFFERS
DBCC FREEPROCCACHE
It will flush the cache memory and rebuild the execution plan next time you will run the stored proc.
msdn.microsoft.com

I think I had the same problem. I removed my parameters from the subqueries. It ran fine after that. Not sure if this is possible in your script but that is what solved it for me.

An answer of Brent Ozar might work, but it returns only active command text by default. For example, it returns WAITFOR DELAY '00:00:05' for query like:
CREATE PROCEDURE spGetChangeNotifications
AS
BEGIN
SET NOCOUNT ON;
DECLARE
#actionType TINYINT;
WHILE #actionType IS NULL
BEGIN
WAITFOR DELAY '00:00:05';
SELECT TOP 1
#actionType = [ActionType]
FROM
TableChangeNotifications;
END;
SELECT
TOP 1000 [RecordID], [Component], [TableName], [ActionType], [Key1], [Key2], [Key3]
FROM
TableChangeNotifications;
END;
How it looks like:
Thus, check the parameter #get_outer_command as described here.
Also, try this one instead(slightly modified procedure from MS Docs):
DECLARE
#sessions TABLE
(
[SPID] INT,
STATUS VARCHAR(MAX),
[Login] VARCHAR(MAX),
[HostName] VARCHAR(MAX),
[BlkBy] VARCHAR(MAX),
[DBName] VARCHAR(MAX),
[Command] VARCHAR(MAX),
[CPUTime] INT,
[DiskIO] INT,
[LastBatch] VARCHAR(MAX),
[ProgramName] VARCHAR(MAX),
[SPID_1] INT,
[REQUESTID] INT
);
INSERT INTO #sessions
EXEC sp_who2;
SELECT
[req].[session_id],
[A].[Login] AS 'login',
[A].[HostName] AS 'hostname',
[req].[start_time],
[cpu_time] AS 'cpu_time_ms',
OBJECT_NAME([st].[objectid], [st].[dbid]) AS 'object_name',
SUBSTRING(REPLACE(REPLACE(SUBSTRING([ST].text, ([req].[statement_start_offset] / 2) + 1, ((CASE [statement_end_offset]
WHEN -1
THEN DATALENGTH([ST].text)
ELSE [req].[statement_end_offset]
END - [req].[statement_start_offset]) / 2) + 1), CHAR(10), ' '), CHAR(13), ' '), 1, 512) AS [statement_text],
[ST].text AS 'full_query_text'
FROM
sys.dm_exec_requests AS req
CROSS APPLY
sys.dm_exec_sql_text(req.sql_handle) AS ST
LEFT JOIN #sessions AS A
ON A.SPID = req.session_id
ORDER BY
[cpu_time] DESC;
How it looks like:
Of course, it's possible to modify code from Brent Ozar answer so it would select a full query text, too, though. Nearly same technique is chosen there(link of code of 18.07.2020 so might change after time):

I had the same problem today and I don't know what causes it but I found a solution. I took the input parameter and saved it into a new parameter, i.e.
declare #parameter2 as x = #parameter
Then i changed the references to the parameter in the queries from #parameter to #parameter2.

Related

How do I chain a VACUUM off of a purge routine running with pg_cron?

Postgres 13.4
I've got some pg_cron jobs set up to periodically delete older records out of log-like files. What I'd like to do is to run VACUUM ANALYZE after performing a purge. Unfortunately, I can't work out how to do this in a stored function. Am I missing a trick? Is a stored procedure more appropriate?
As an example, here's one of my purge routines
CREATE OR REPLACE FUNCTION dba.purge_event_log (
retain_days_in integer_positive default 14)
RETURNS int4
AS $BODY$
WITH -- Use a CTE so that we've got a way of returning the count easily.
deleted AS (
-- Normal-looking code for this requires a literal:
-- where your_dts < now() - INTERVAL '14 days'
-- Don't want to use a literal, SQL injection, etc.
-- Instead, using a interval constructor to achieve the same result:
DELETE
FROM dba.event_log
WHERE dts < now() - make_interval (days => $1)
RETURNING *
),
----------------------------------------
-- Save details to a custom log table
----------------------------------------
logit AS (
insert into dba.event_log (name, details)
values ('purge_event_log(' || retain_days_in::text || ')',
'count = ' || (select count(*)::text from deleted)
)
)
----------------------------------------
-- Return result count
----------------------------------------
select count(*) from deleted;
$BODY$
LANGUAGE sql;
COMMENT ON FUNCTION dba.purge_event_log (integer_positive) IS
'Delete dba.event_log records older than the day count passed in, with a default retention period of 14 days.';
The truth is, I don't really care about the count(*) result from this routine, in this case. But I might want a result and an additional action in some other, similar context. As you can see, the routine deletes records, uses a CTE to insert a report into another table, and then returns a result. No matter what, I figure this example is a good way to get me head around the alternatives and options in stored functions. The main thing I want to achieve here is to delete records, and then run maintenance. if this is an awkward fit for a stored function or procedure, I could write out an entry to a vacuum_list table with the table name, and have another job to run though that list.
If there's a smarter way to approach vacuum without the extra, I'm of course interested in that. But I'm also interested in understanding the limits on what operationa you can combine in PL/PgSQL routines.
Pavel Stehule' answer is correct and complete. I decided to follow-up a bit here as I like to dig in on bugs in my code, behaviors in Postgres, etc. to get a better sense of what I'm dealing with. I'm including some notes below for anyone who finds them of use.
COMMAND cannot be executed...
The reference to "VACUUM cannot be executed inside a transaction block" gave me a better way to search the docs for similarly restricted commands. The information below probably doesn't cover everything, but it's a start.
Command Limitation
CREATE DATABASE
ALTER DATABASE If creating a new table space.
DROP DATABASE
CLUSTER Without any parameters.
CREATE TABLESPACE
DROP TABLESPACE
REINDEX All in system catalogs, database, or schema.
CREATE SUBSCRIPTION When creating a replication slot (the default behavior.)
ALTER SUBSCRIPTION With refresh option as true.
DROP SUBSCRIPTION If the subscription is associated with a replication slot.
COMMIT PREPARED
ROLLBACK PREPARED
DISCARD ALL
VACUUM
The accepted answer indicates that the limitation has nothing to do with the specific server-side language used. I've just come across an older thread that has some excellent explanations and links for stored functions and transactions:
Do stored procedures run in database transaction in Postgres?
Sample Code
I also wondered about stored procedures, as they're allowed to control transactions. I tried them out in PG 13 and, no, the code is treated like a stored function, down to the error messages.
For anyone that goes in for this sort of thing, here are the "hello world" samples of sQL and PL/PgSQL stored functions and procedures to test out how VACCUM behaves in these cases. Spoiler: It doesn't work, as advertised.
SQL Function
/*
select * from dba.vacuum_sql_function();
Fails:
ERROR: VACUUM cannot be executed from a function
CONTEXT: SQL function "vacuum_sql_function" statement 1. 0.000 seconds. (Line 13).
*/
DROP FUNCTION IF EXISTS dba.vacuum_sql_function();
CREATE FUNCTION dba.vacuum_sql_function()
RETURNS VOID
LANGUAGE sql
AS $sql_code$
VACUUM ANALYZE activity;
$sql_code$;
select * from dba.vacuum_sql_function(); -- Fails.
PL/PgSQL Function
/*
select * from dba.vacuum_plpgsql_function();
Fails:
ERROR: VACUUM cannot be executed from a function
CONTEXT: SQL statement "VACUUM ANALYZE activity"
PL/pgSQL function vacuum_plpgsql_function() line 4 at SQL statement. 0.000 seconds. (Line 22).
*/
DROP FUNCTION IF EXISTS dba.vacuum_plpgsql_function();
CREATE FUNCTION dba.vacuum_plpgsql_function()
RETURNS VOID
LANGUAGE plpgsql
AS $plpgsql_code$
BEGIN
VACUUM ANALYZE activity;
END
$plpgsql_code$;
select * from dba.vacuum_plpgsql_function();
SQL Procedure
/*
call dba.vacuum_sql_procedure();
ERROR: VACUUM cannot be executed from a function
CONTEXT: SQL function "vacuum_sql_procedure" statement 1. 0.000 seconds. (Line 20).
*/
DROP PROCEDURE IF EXISTS dba.vacuum_sql_procedure();
CREATE PROCEDURE dba.vacuum_sql_procedure()
LANGUAGE SQL
AS $sql_code$
VACUUM ANALYZE activity;
$sql_code$;
call dba.vacuum_sql_procedure();
PL/PgSQL Procedure
/*
call dba.vacuum_plpgsql_procedure();
ERROR: VACUUM cannot be executed from a function
CONTEXT: SQL statement "VACUUM ANALYZE activity"
PL/pgSQL function vacuum_plpgsql_procedure() line 4 at SQL statement. 0.000 seconds. (Line 23).
*/
DROP PROCEDURE IF EXISTS dba.vacuum_plpgsql_procedure();
CREATE PROCEDURE dba.vacuum_plpgsql_procedure()
LANGUAGE plpgsql
AS $plpgsql_code$
BEGIN
VACUUM ANALYZE activity;
END
$plpgsql_code$;
call dba.vacuum_plpgsql_procedure();
Other Options
Plenty. As I understand it, VACUUM, and a handful of other commands, are not supported in server-side code running within Postgres. Therefore, you code needs to start from somewhere else. That can be:
Whatever cron you've got in your server's OS.
Any exteral client you like.
pg_cron.
As we're deployed on RDS, those last two options are where I'll look. And there's one more:
Let AUTOVACCUM and an occasional VACCUM do their thing.
That's pretty easy to do, and seems to work fine for the bulk of our needs.
Another Idea
If you do want a bit more control and some custom logging, I'm imagining a table like this:
CREATE TABLE IF NOT EXISTS dba.vacuum_list (
database_name text,
schema_name text,
table_name text,
run boolean,
run_analyze boolean,
run_full boolean,
last_run_dts timestamp)
ALTER TABLE dba.vacuum_list ADD CONSTRAINT
vacuum_list_pk
PRIMARY KEY (database_name, schema_name, table_name);
That's just a sketch. The idea is like this:
You INSERT into vacuum_list when a table needs some vacuuming, at least as far as you're concerned.
In my case, that would be an UPSERT as I don't need a full log-like table, just a single row per table of interest with the last outcome and/or pending state.
Periodically, a remote client, etc. connects, reads the table, and executes each specified VACUUM, according to the options specified in the record.
The external client updates the row with the last run timestamp, and whatever else you're including in the row.
Optionally, you could include fields for duration and change in relation size pre:post vacuuming.
That last option is what I'm interested in. None of our VACUUM calls were working for quite some time as there was a months-old dead connection from something sever-side. VACUUM appears to run fine, in such a case, it just can't delete a whole lot of rows. (Because of the super old "open" transaction ID, visibility maps, etc.) The only way to see this sort of thing seems to be to VACUUM VERBOSE and study the output. Or to record vacuum time and, more important, relation size change to flag cases where nothing seems to happen, when it seems like it should.
VACUUM is "top level" command. It cannot be executed from PL/pgSQL ever or from any other PL.

At what point are DB2 declared global temporary tables automatically deleted...?

When are DB2 declared global temporary tables 'cleaned up' and automatically deleted by the system...? This is for DB2 on AS400 v7r3m0, with DBeaver 5.2.5 as the dev client, and MS-Access 2007 for packaged apps for the end-users.
Today I started experimenting with a DGTT, thanks to this answer. So far I'm pleased with the functionality, although I did find our more recent system version has the WITH DATA option, which is an obvious advantage.
Everything is working, but at times I receive this error:
SQL Error [42710]: [SQL0601] NEW_PKG_SHEETS_DATA in QTEMP type *FILE already exists.
The meaning of the error is obvious, but the timing is not. When I started today, I could run the query multiple times, and the error didn't occur. It seemed as if the system was cleaning up and deleting it, which is just what I was looking for. But then the error started and now it's happening with more frequency.
If I make strategic use of DROP TABLE, this resolves the error, unless the table doesn't exist, in which case I get another error. I can also disconnect/reconnect to the server from my SQL dev client, as I would expect, since that would definitely drop the session.
This IBM article about DGTTs speaks much of sessions, but not many specifics. And this article is possibly the longest command syntax I've yet encountered in the IBM documentation. I got through it, but it didn't answer the question of what decided when a DGTT is deleted.
So I would like to ask:
What are the boundaries of a session..?
I'm thinking this is probably defined by the environment in my SQL client..?
I guess the best/safest thing to do is use DROP TABLE as needed..?
Does any one have any tips, tricks, or pointers they could share..?
Below is the SQL that I'm developing. For brevity, I've excluded chunks of the WITH-AS and SELECT statements:
DROP TABLE SESSION.NEW_PKG_SHEETS ;
DECLARE GLOBAL TEMPORARY TABLE SESSION.NEW_PKG_SHEETS_DATA
AS ( WITH FIRSTDAY AS (SELECT (YEAR(CURDATE() - 4 MONTHS) * 10000) +
(MONTH(CURDATE() - 4 MONTHS) * 100) AS DATEISO
FROM SYSIBM.SYSDUMMY1
-- <VARIETY OF ADDITIONAL CTE CLAUSES>
-- <SELECT STATEMENT BELOW IS A BIT LONGER>
SELECT DAACCT AS DAACCT,
DAIDAT AS DAIDAT,
DAINV# AS DAINV,
CAST(DAITEM AS NUMERIC(6)) AS DAPACK,
CAST(0 AS NUMERIC(14)) AS UPCNUM,
DAQTY AS DAQTY
FROM DAILYTRANS
AND DAIDAT >= (SELECT DATEISO+000 FROM FIRSTDAY) -- 1ST DAY FOUR MONTHS AGO
AND DAIDAT <= (SELECT DATEISO+399 FROM FIRSTDAY) -- LAST DAY OF LAST MONTH
) WITH DATA ;
DROP TABLE SESSION.NEW_PKG_SHEETS ;
The DGTT will only get cleaned automatically up by Db2 when the connection ends successfully (connect reset or equivalent according to whatever interface to Db2 is being used ).
For both Db2 for i and Db2-LUW, consider using the WITH REPLACE clause for the DECLARE GLOBAL TEMPORARY TABLE statement. That will ensure you don't need to explicitly drop the DGTT if the session remains open but the code needs the table to be replaced at next execution whether or not the DGTT already exists.
Using that WITH REPLACE clause means you do not need to worry about issuing a DROP statement for the DGTT, unless you really want to issue a drop.
Sometimes sessions may get re-used, or a close/disconnect might not happen or might not complete, or more likely a workstation performs a retry, and in those cases the WITH REPLACE can be essential for easily avoiding runtime errors.
Note that Db2 for Z/OS (at v12) does not offer the WITH REPLACE clause for DGTT, but has instead an optional syntax on commit drop table (but this is not documented for Db2-for-i and Db2-LUW).

Postgresql insert stopped working, duplicate key value violations

About 8 months ago I used a suggestion to set up a holding table, then push to the formal table and prevent duplicate entries, per this post: Best way to prevent duplicate data on copy csv postgresql
It's been working very nicely, but today I noticed some errors and gaps in the data.
Here's my insert statement:
And here's how the index is set up:
And here's an example of the error I'm getting, although on the next chron insert, it went through.
Here's it going through fine:
I haven't noted any large changes in the data incoming. Here's what the data looks like that's coming in now:
In summary, I've noticed recent oddities with the insert statements, and success is erratic, resulting in large data gaps in the database. Thanks for any help, and I'm happy to provide more details, but I wanted to see if my information sounds like something someone else has already dealt with.
Thanks very much for any help,
S
As Gordon pointed out in his answer to your previous question, this approach only works if you have exclusive access to the table. There is a delay between the existence check and the insert itself, and if another process modifies the table during this window, you may end up with duplicates.
If you're on Postgres 9.5+, the best approach is to skip the existence check and simply use an INSERT ... ON CONFLICT DO NOTHING statement.
On earlier versions, the simplest solution (if you can afford to do so) would be to lock the table for the duration of the import. Otherwise, you can emulate ON CONFLICT DO NOTHING (albeit less efficiently) using a loop and an exception handler:
DO $$
DECLARE r RECORD;
BEGIN
FOR r IN (SELECT * FROM holding) LOOP
BEGIN
INSERT INTO ltg_data (pulsecount, intensity, time, lon, lat, ltg_geom)
VALUES (r.pulsecount, r.intensity, r.time, r.lon, r.lat, r.ltg_geom);
EXCEPTION WHEN unique_violation THEN
END;
END LOOP;
END
$$
As an aside, DELETE FROM holding is time-consuming and probably unnecessary. Create your staging table as TEMP, and it will be cleaned up automatically at the end of your session. You can easily build a table which matches the structure of your import target:
CREATE TEMP TABLE holding (LIKE ltg_data);

SQL Server temp tables via MS Access

Well I've been using #temp tables in standard T-SQL coding for years and thought I understood them.
However, I've been dragged into a project based in MS Access, utilizing pass-through queries, and found something that has really got me puzzled.
Though maybe it's the inner workings of Access that has me fooled !?
Here we go : Under normal usage, I understand the if I create a temp table in a Sproc, it's scope ends with the end of the SProc, and is dropped by default.
In the Access example, I found it was possible to do this in one Query:
select top(10) * into #myTemp from dbo.myTable
And then this in second separate query:
select * from #myTemp
How is this possible ?
If a temp table dies with the current session, does this mean that Access keeps a single session open, and uses that session for all Queries executed ?
Or has my fundamental understanding of scope been wrong all this time ?
Hope someone out there can help clarify what is occurring under the hood !?
Many Thanks
I found this answer of a kind of similar question:
Temp table is stored in tempdb until the connection is dropped (or in the case of a global temp tables when the last connection using it is dropped). You can also (and it is a good proctice to do so) manually drop the table when you are finished using it with a drop table statement.
I hope this helps out.

Multiple prepared statements disrupt a transaction using DBD::Sybase

In my Perl script, I use DBD::Sybase (via DBI module) to connect to a SQL Server 2008. The base program as below runs without problem:
use DBI;
# assign values to $host, $usr, $pwd
my $dbh = DBI->connect("dbi:Sybase:$host", $usr, $pwd);
$dbh->do("BEGIN TRAN tr1");
my $update = $dbh->prepare("UPDATE mytable SET qty = ? where name = ?");
$update->execute(100, 'apple');
$dbh->do("END TRAN tr1");
however, if I insert one more prepare statement right before the existing prepare statement, to have the program look like:
...
my $insert = $dbh->prepare("INSERT INTO mytable (name, qty) VALUES (?, ?)");
my $update = $dbh->prepare("UPDATE mytable SET qty = ? where name = ?");
...
and the rest is all the same, then when I run it, I got:
DBD::Sybase::db do failed: Server message number=3902 severity=16 state=1 line=1 server=xxx text=The COMMIT TRANSACTION request has no corresponding BEGIN TRANSACTION.
So looks like the additional prepare statement somehow disrupted the entire transaction flow. I had been running the same code via the DBD::ODBC driver with no problem against a SQL SERVER 2005. (But my firm upgraded to 2008 and I had to use the DBD::Sybase to get around some other problems.)
Any help / suggestion on how to resolve this issue would be much appreciated. In particular, using a different db handle for the other prepare is not a desired solution since that will beat the purpose of having them in a single transaction.
UPDATE: Turns out if I execute at least once on the additional insert, then the program is again run fine. So looks like every prepared statement needs to be run under Sybase. But that isn't a requirement with ODBC and isn't a reasonable requirement in general. Anyway to get around it?
You are learning perl AND Sybase basics and making several incorrect conclusions.
Forget about what it does under ODBC for a moment. ODBC most probably has AUTOCOMMIT turned on, and thus you have no transaction control whatsoever. (Why anyone would use ODBC when the DBD:: supports DB-Lib and CT-Lib is beyond me, but that's a separate story.)
Re: "So looks like every prepared statement needs to be run under Sybase."
Rawheiser is correct. What exactly do you expect to achieve by preparing a batch but performing a Do instead ? Where else do you expect to execute the batch prepared under Sybase, other than under Sybase?
Do vs prepare/execute are quite different. prepare/execute for Sybase works just fine in millions of programs. you just have to learn what it does, not what you think it should do. prepare let's you load a batch, a block of commands terminated by GO in the normal Sybase sense. Execute executes the prepared batch (supplies the GO and sends the batch to the server), and captures whatever is returned (according to whatever array/variables you have set).
Do is immediate, single command, with no prepare. A prepare+execute combined.
Performing only single-statement do's, and only dynamic SQL, simply because that's all that you could get to work, is very limiting and quite unnecessary.
You currently have:
Prepare:
UPDATE
Execute (100)
ExecuteImmediate(Do):
COMMIT TRAN
So of course, there is no BEGIN TRAN. (The first "do" executed, the BEGIN TRAN is gone)
I think what you want (intended originally) is this. Forget the 'do':
Prepare:
BEGIN TRAN
UPDATE
COMMIT TRAN
Execute (100)
Then change it to:
BEGIN TRAN
INSERT
UPDATE
COMMIT TRAN
Execute (100)
Your $update and $insert will confuse you (you're executing a multi-statement batch, right ?not a isolated single command in the middle of a prepare batch). If you get rid of them, and think in terms of $execute [whatever you have prepared in the batch], it might help you to understand the problem better.
Do not form conclusions until you have all the above working as intended.
And read up on BEGIN/COMMIT TRAN.
Last, What exactly is a "END TRAN" ? I do not think the code block you have posted is real.
Don't dynamically create SQL, it is dangerous (sql injection).
You should be able to prepare multiple inserts/updates and your link to the DBI documentation does not say you cannot, it says some drivers may not be able to tell you much about a statement which is ONLY prepared.
I'd post a failing example with error to the dbi-users list for comment as the DBD::Sybase maintainer hangs out there (see dbi.perl.org).
Turns out that DBI's prepare method is not quite portable across various database drivers as noted here. For the Sybase driver, it is most likely that prepare is not working as intended. One way to tell is that after running prepare, the variable $insert->{NUM_OF_FIELDS} is undefined.
To get around the problem, do one of the following:
1) do not prepare anything. Just dynamically construct the statement in text string and run $dbh->do($stmt), or
2) run finish on all outstanding statement handles (under that database handle) before running COMMIT TRAN. I personally prefer this way much better.