Monday, October 9, 2017

How To Shrink Foxhound

Question: How do I shrink and reorganize the Foxhound 4 database?

Answer: Foxhound uses SQL Anywhere for its database, and the only way to shrink the size of a SQL Anywhere database file is to unload, recreate and reload the file from scratch. The good news is, that whole process is automatic when you reinstall Foxhound on an existing installation:

  • First, Foxhound creates a new, empty copy of the database file.

  • Second, Foxhound finds the old database file,

  • then it asks you how much of the old data you want to copy to the new database: some, all, none, just the options.

  • Finally, Foxhound copies and loads the data, using just enough space to hold it all.
In this context, "shrink" means "don't copy the empty space, and don't allocate space for data that isn't copied", and "reorganize" means the copy process implicitly organizes the data as it is loaded.

Here's the step-by-step process:

Step 1: Start Foxhound 4

...if it isn't running already, that is.

In this example, the Foxhound 4 database has grown quite large:
 Directory of C:\ProgramData\RisingRoad\Foxhound4

10/02/2017  07:31 PM    37,521,506,304 foxhound4.db

Step 2: Decide How Much Data To Copy

Let's say you want to save half the data, thus shrinking the file by 50%.

Foxhound doesn't understand "half", but it does understand "only copy samples recorded after yyyymmdd".

To convert "half" into "yyyymmdd", start ISQL and run this query:
All Programs - Foxhound4 - Tools - Adhoc Query Foxhound Database via ISQL

SELECT CAST ( SYSTAB.count * 0.50 AS INTEGER ) AS half
  FROM SYSTAB
 WHERE SYSTAB.table_name = 'rroad_sample_set';

       half 
----------- 
     867871 

-- The space used by the foxhound4.db file is more-or-less determined by the number 
-- of rows in the rroad_sample_set table, which contains one row for each sample
-- sample recorded by the Foxhound Monitor.

-- In other words, the "half" calculated here is "half the samples", not "half the bytes".
Now use "half" as the START AT value in this query:
SELECT TOP 1 START AT 867871
       DATEFORMAT ( sample_header.sample_finished_at, 'yyyymmdd' ) AS yyyymmdd
  FROM sample_header
 ORDER BY sample_header.sample_set_number;

yyyymmdd 
-------- 
20170808 

-- The sample_finished_at column is used to turn the row number 867871 into a yyyymmdd date.
-- The sample_header view is used because Foxhound only allows SELECT statements on the 
-- adhoc query views (sample_header), not the underlying tables (rroad_sample_set).

Step 3: Run Foxhound's "Unsetup"

You don't have to stop Foxhound to reinstall the software, but you do have to run the special "unsetup" process to prepare for the reinstallation:
All Programs - Foxhound4 - Tools - Unsetup Foxhound

-- The "unsetup" process gets rid of the Windows shortcuts and other items, 
-- but it leaves the Foxhound database alone... and even leaves it running.

Step 4: Run The Foxhound 4 Setup Up To The PLEASE READ THIS

In this example, the new Foxhound 4 build 4740 is used... run it all the way to the PLEASE READ THIS window:
Foxhound-Version-4-0-4740-setup.exe


Step 5: Enter The FOXHOUND4UPGRADE=yyyymmdd Value

You can copy and paste the FOXHOUND4UPGRADE=yyyymmdd value, and press Enter twice to continue:

Step 6: Wait... and Wait... and Wait Some More

It takes quite a while for 35G of data to be copied and loaded, even half of 35G.


Eventually, the upgrade process will finish, and shut down, and Foxhound 4 will be restarted with the newer, smaller, reorganized database:
 Directory of C:\ProgramData\RisingRoad\Foxhound4

10/06/2017  07:25 PM    21,720,817,664 foxhound4.db
That's 42% smaller that before... not exactly half, but it meets the "more-or-less" standard :)

Thursday, October 5, 2017

New Foxhound 4 Build 4740

A new build of Foxhound 4 is available here.

  • You can use it to upgrade an existing copy of Foxhound 4 for free.

  • You can also use it to install a new copy of Foxhound 4, or

  • to upgrade an existing copy of Foxhound 1, 2 or 3.
If you are already using Foxhound 4, here's why you should consider upgrading to build 4740:
  • Build 4740 "rolls up" all five patches that were previously released for the original Foxhound 4 build 4729.

  • Several performance problems have been fixed, making it less likely that Foxhound will become unresponsive when the database grows very large.

  • If you want to use the "Reinstall Foxhound" method to reorganize and shrink the Foxhound database, the new build 4740 will let you do that... unlike the previous Patch 10 which made it impossible to preserve any data if you subsequently reinstalled the original build 4729.

  • Foxhound's own purge process has been improved, making it more likely the purge will keep database growth under control... and this fix was not included in the previous patches.
On the other hand...
  • If you don't need any of changes in build 4740, there's no need to upgrade.

  • If you only need one or two of the fixes, applying a patch to to build 4729 is a lot quicker than upgrading to build 4740.
Either way, here's a tip...
Tip: Set the Purge Speed to 10 Fastest on section 6. Purge of the Foxhound Options page.

Do this whether or not you upgrade to build 4740, if your Foxhound database is growing rapidly.


Sunday, September 24, 2017

Latest Foxhound 4 Patches

Two new patches have been posted recently on the Foxhound 4 page:

[ Download Patch 3 ]  Fix "unable to load dll" messages:
 Could not load dynamic library
 Unable to load either rroad4.dll or rroad464.dll
[ Patch 3 Readme ]
[ Download Patch 4 ]  Fix Display Schema for billion-row tables:
 Value nnn out of range for destination
 [ Patch 4 Readme ]
[ Download Patch 5 ]   NEW Prevent Foxhound from becoming unresponsive
 when the Monitor cannot connect to a network target database
 and an endless loop of bogus error messages appears:
  Attempt to reload definition for event "rroad_monitor_sample_loop"
  failed due to concurrent event execution.
 [ Patch 5 Readme ]
[ Download Patch 8 ]  Prevent Foxhound from becoming unresponsive
 when Ping-Only Sampling is stopped and restarted, or
 when full sampling is attempted for an arbiter server.
 [ Patch 8 Readme ]
[ Download Patch 10 ]   NEW Prevent Foxhound from becoming unresponsive
 on startup, especially when the Foxhound database is very large.
 [ Patch 10 Readme ]

The two new patches are more likely to help Foxhound users than the others:

Patch 5 will help folks who are using recent builds of SQL Anywhere 16 and 17 to run Foxhound. In particular, it implements a workaround for a bug behavior change introduced after SQL Anywhere 16.0.0.2193 and SQL Anywhere 17.0.0.1359.

Patch 10 will help anyone with a Foxhound database that has grown into the multi-gigabyte range by decreasing the time it takes for Foxhound to start up. For example, here are the test results for a 1.4G database:
Before applying Patch 10:  Foxhound startup time: 10.8s
After:                     Foxhound startup time:  1.1s
That's a factor of 10, but... you might not notice a difference of only 9 seconds.

However, you will notice the effect with a larger Foxhound database. For example, here's what happened with a 35G database:
Before applying Patch 10:  Foxhound startup time: 7m 17s ...plus 30 minutes more for the Foxhound Menu to appear!
After:                     Foxhound startup time: 4.6s
If you see this message every time you start Foxhound 4, you probably need to apply Patch 10:
Foxhound was not available yet, probably because the Foxhound engine had not finished starting up.
It may be ready now, so click here to try again.

If the situation persists contact RisingRoad.
If that message goes away after a long while, replaced by a blank page, and you still don't see the Foxhound menu appear, you definitely need Patch 10... especially if Foxhound's instance of dbsrv16.exe is using up this much CPU:

The "Foxhound startup time: 7m 17s" only tells part of the story in the example above... after Foxhound finished starting up, it took a further 30 minutes before the menu page appeared!

How Patch 10 Was Created...
Step 1: Find the Queries From Hell


The first step was to start SQL Anywhere's Procedure Profiler as soon as Foxhound started executing, by adding this statement to DatabaseStart event script:
CALL sa_server_option ( 'ProcedureProfiling', 'YES' );
Then, after the Foxhound menu finally appeared, this dbisql query showed which SQL statements took the longest to execute:
SELECT * 
  FROM sa_procedure_profile()
 where millisecs >= 100
 ORDER BY object_name, line_num;
Here's the row that shows line 102 in the rroad_monitor_list_html stored procedure took 473,628 milliseconds to run, which is 8 minutes:
object_type, object_name,               owner_name, table_name, line_num, executions, millisecs, percentage,       foreign_owner, foreign_table
'F',         'rroad_monitor_list_html', 'DBA',      ,           102,      1,          473628,    99.9972553051784, ,
Here's how to find line 102 that stored procedure:
Tip: You can't use Sybase Central to view the source code because the version it displays (SYSPROCEDURE.source) isn't the version that executes (SYSPROCEDURE.proc_defn), so you have use an Old School query.
UNLOAD SELECT proc_defn 
         FROM SYSPROCEDURE 
        WHERE proc_name = 'rroad_monitor_list_html' 
    TO 'C:\temp2\\temp_rroad_monitor_list_html.txt' 
    DELIMITED BY '' ESCAPES OFF HEXADECIMAL OFF QUOTES OFF;
Here's the Query From Hell at line 102:
select "rroad_alert_union"."sampling_id" as "sampling_id",
  "COUNT"(distinct "rroad_alert_union"."alert_number") as "active_alert_count",
  "LIST"(
  distinct "STRING"(
  '#',
  "rroad_alert_union"."alert_number"),
  ', ' order by
  "rroad_alert_union"."alert_number" asc) as "active_alert_number_list"
  into local temporary table "active_alert_count"
  from "rroad_alert_union"
  where "rroad_alert_union"."record_type" = 'Alert'
  and "rroad_alert_union"."alert_is_clear_or_cancelled" = 'N'
  group by "rroad_alert_union"."sampling_id";

Step 2: Study the Query Plans From Hell


The next step was to copy and paste the Query From Hell into dbisql and click on Tools - Plan Viewer.

In this case, the "Main Query" didn't look too bad, just one table scan for 77K rows, but the subquery called "SubQ 5" was truly breathtaking... a 77K table scan repeated 77K times!

SubQ 5 was found lurking down inside a view, and there was no apparent way to make it run faster; no indexes helped, nor did a materialized view:
LEFT OUTER JOIN ( SELECT *, 
                         'Y' AS alert_is_clear
                    FROM rroad_all_clear 
                ) AS rroad_all_clear
             ON rroad_all_clear.sampling_id        = rroad_alert.sampling_id
            AND rroad_all_clear.alert_number       = rroad_alert.alert_number
            AND rroad_all_clear.alert_in_effect_at = rroad_alert.alert_in_effect_at 

The only solution was eliminate SubQ 5 altogether by adding a new derived (redundant) column to one of the tables.

Step 3: Fix the Queries From Hell


Patch 10 actually introduces two new derived columns, for two separate Queries From Hell; here's what the code looks like.

The first derived column...
ALTER TABLE rroad_alert 
   ADD alert_is_clear_or_cancelled   VARCHAR ( 1 ) NOT NULL DEFAULT 'N';
allowed this slow query code...
CREATE VIEW rroad_alert_union AS
SELECT CAST ( 'Alert' AS VARCHAR ( 20 ) )       AS record_type,
       ...
       COALESCE ( 
          rroad_all_clear.alert_is_clear, 
          rroad_alert_cancelled.alert_is_cancelled, 
          'N' )                                 AS alert_is_clear_or_cancelled
  FROM rroad_alert
          INNER JOIN rroad_sampling_options
                  ON rroad_sampling_options.sampling_id = rroad_alert.sampling_id
          LEFT OUTER JOIN ( SELECT *, 
                                   'Y' AS alert_is_clear
                              FROM rroad_all_clear 
                          ) AS rroad_all_clear
                       ON rroad_all_clear.sampling_id        = rroad_alert.sampling_id
                      AND rroad_all_clear.alert_number       = rroad_alert.alert_number
                      AND rroad_all_clear.alert_in_effect_at = rroad_alert.alert_in_effect_at 
          LEFT OUTER JOIN ( SELECT *,  
                                   'Y' AS alert_is_cancelled
                              FROM rroad_alert_cancelled 
                          ) AS rroad_alert_cancelled
                       ON rroad_alert_cancelled.sampling_id        =  rroad_alert.sampling_id
                      AND rroad_alert_cancelled.alert_number       =  rroad_alert.alert_number
                      AND rroad_alert_cancelled.alert_in_effect_at = rroad_alert.alert_in_effect_at
to be simplified (and speeded up) by eliminating both outer join subqueries (including SUBQ 5)...
CREATE VIEW rroad_alert_union AS
SELECT CAST ( 'Alert' AS VARCHAR ( 20 ) )       AS record_type,
       ...
       rroad_alert.alert_is_clear_or_cancelled  AS alert_is_clear_or_cancelled
  FROM rroad_alert
          INNER JOIN rroad_sampling_options
                  ON rroad_sampling_options.sampling_id = rroad_alert.sampling_id
Two new triggers were required to maintain the new column:
CREATE OR REPLACE TRIGGER tri_rroad_all_clear
   AFTER INSERT ON rroad_all_clear
   REFERENCING NEW AS new_rroad_all_clear
   FOR EACH ROW
BEGIN
   UPDATE rroad_alert
      SET rroad_alert.alert_is_clear_or_cancelled = 'Y'
    WHERE rroad_alert.sampling_id        = new_rroad_all_clear.sampling_id
      AND rroad_alert.alert_number       = new_rroad_all_clear.alert_number
      AND rroad_alert.alert_in_effect_at = new_rroad_all_clear.alert_in_effect_at;
END;

CREATE OR REPLACE TRIGGER tri_rroad_alert_cancelled
   AFTER INSERT ON rroad_alert_cancelled
   REFERENCING NEW AS new_rroad_alert_cancelled
   FOR EACH ROW
BEGIN
   UPDATE rroad_alert
      SET rroad_alert.alert_is_clear_or_cancelled = 'Y'
    WHERE rroad_alert.sampling_id        = new_rroad_alert_cancelled.sampling_id
      AND rroad_alert.alert_number       = new_rroad_alert_cancelled.alert_number
      AND rroad_alert.alert_in_effect_at = new_rroad_alert_cancelled.alert_in_effect_at;
END;
Patch 10 contains another derived column...
ALTER TABLE rroad_sampling_options 
   ADD latest_sample_set_number      UNSIGNED BIGINT NOT NULL DEFAULT 0;
which allowed this slow query code...
LEFT OUTER JOIN ( SELECT rroad_group_1_property_pivot.*
                    FROM rroad_group_1_property_pivot
                         INNER JOIN ( SELECT rroad_group_1_property_pivot.sampling_id               AS sampling_id,
                                             MAX ( rroad_group_1_property_pivot.sample_set_number ) AS sample_set_number
                                        FROM rroad_group_1_property_pivot
                                       GROUP BY rroad_group_1_property_pivot.sampling_id 
                                    ) AS latest_primary_key
                                 ON latest_primary_key.sampling_id       = rroad_group_1_property_pivot.sampling_id
                                AND latest_primary_key.sample_set_number = rroad_group_1_property_pivot.sample_set_number
                 ) AS rroad_group_1_property_pivot
             ON rroad_group_1_property_pivot.sampling_id = rroad_sampling_options.sampling_id
to be simplified (and speeded up) by eliminating the complex inner join subquery:
LEFT OUTER JOIN rroad_group_1_property_pivot
             ON rroad_group_1_property_pivot.sampling_id       = rroad_sampling_options.sampling_id
            AND rroad_group_1_property_pivot.sample_set_number = rroad_sampling_options.latest_sample_set_number
Two existing triggers had to be modified to maintain the second derived column:
ALTER TRIGGER tri_rroad_sample_set
   AFTER INSERT ON rroad_sample_set
   REFERENCING NEW AS new_rroad_sample_set
   FOR EACH ROW
BEGIN
...
UPDATE rroad_sampling_options
   SET rroad_sampling_options.latest_sample_set_number = new_rroad_sample_set.sample_set_number
 WHERE rroad_sampling_options.sampling_id              = new_rroad_sample_set.sampling_id
   AND rroad_sampling_options.latest_sample_set_number < new_rroad_sample_set.sample_set_number;
...
END; -- tri_rroad_sample_set

ALTER TRIGGER tru_rroad_sample_set
   BEFORE UPDATE OF sample_finished_at ON rroad_sample_set
   REFERENCING OLD AS old_rroad_sample_set NEW AS new_rroad_sample_set
   FOR EACH ROW
BEGIN
...
IF new_rroad_sample_set.sample_set_number <> old_rroad_sample_set.sample_set_number THEN

   UPDATE rroad_sampling_options
      SET rroad_sampling_options.latest_sample_set_number = new_rroad_sample_set.sample_set_number
    WHERE rroad_sampling_options.sampling_id              = new_rroad_sample_set.sampling_id
      AND rroad_sampling_options.latest_sample_set_number < new_rroad_sample_set.sample_set_number;

END IF;
...
END; -- tru_rroad_sample_set

Sunday, July 23, 2017

Characteristic Errors, Revision 6

A characteristic error is a mistake so easy to make it appears you're actively encouraged to make it by the very nature of the software you're using.

Here's the latest entry...

  1. SQL Anywhere: Forgetting to code AUTOSTOP OFF on a START DATABASE statement, then wondering why the database immediately stopped after starting.
Here's the full list...
  1. SQL: Seeing too little data, or no data at all, because a predicate in the WHERE clause effectively turned your OUTER JOIN into an INNER JOIN.

  2. SQL: Seeing too much data because a missing predicate in the WHERE clause effectively turned your INNER JOIN into a CROSS JOIN.

  3. SQL: Getting the wrong COUNT() or SUM() because you forgot to code WHERE ... IS NOT NULL, or you *did* code it when you shouldn't have.

  4. SQL: Getting the wrong answer because you forgot that, in general, NULL values [cough] suck.

  5. SQL Anywhere: Not seeing MESSAGE output because you forgot to run SET TEMPORARY OPTION DEBUG_MESSAGES = 'ON';

  6. SQL Anywhere: Not seeing any data because you forgot ON COMMIT PRESERVE ROWS or NOT TRANSACTIONAL.

  7. SQL Anywhere: Coding ENDIF where END IF was required, or vice versa (before Version 11).

  8. SQL Anywhere: Connecting to the wrong server because you forgot DOBROAD=NONE (before Version 12).

  9. SQL Anywhere: Forgetting the asterisk in SELECT TOP 10 FROM ...

  10. SQL Anywhere: Coding IF NOT VAREXISTS ( 'x' ) THEN ... instead of IF VAREXISTS ( 'x' ) = 0 THEN ...

  11. SQL Anywhere: Coding the wrong magic numbers 1, 2, 3, ... in the get_value() and set_value() calls in an EXTERNAL C DLL function.

  12. SQL Anywhere: Getting proxy table ODBC errors because the engine's running as a service and you've set up a User DSN instead of System DSN.

  13. SQL Anywhere: Getting file-related errors because the file specifications are relative to the server rather than the client.

  14. SQL Anywhere: Getting file-related errors because the engine's running as a service without the necessary permissions.

  15. SQL Anywhere: Coding CREATE TRIGGER IF NOT EXISTS instead of CREATE OR REPLACE TRIGGER, or vice versa for CREATE TABLE (in 11.0.1 or later).

  16. SQL Anywhere: Getting integer arithmetic when you wanted fractional parts because you forgot to CAST.

  17. Stored procedure debugger: Setting it to watch a specific user id other than the one you're using to test your code.

  18. Sybase Central: Setting it to display objects for owner names other than the one you're interested in.

  19. Copy and paste: Forgetting to edit after pasting; e.g., Copy and paste SET @continue = 'Y' into the body of a WHILE loop and then forgetting to change it to 'N'.

  20. MobiLink: Forgetting to call ml_add_column for any of the columns you're trying to synchronize, thus guaranteeing yourself a "Sassen Frassen Fricken Fracken!" moment when you run the first test.

  21. MobiLink: Forgetting to call ml_add_[various] with the NULL parameter to delete old ml_[whatever] rows, thus ending up with thousands of orphan system table rows in the consolidated database.

  22. OLAP Windowing: Coding the wrong combination of ASC and DESC in an inner OVER ORDER BY clause and the outer SELECT ORDER BY: different when they should be the same, the same when they should be different, or some other variation of "wrong combination"...
                SELECT older_sample_set.sample_set_number
                  INTO @20_older_sample_set_number
                  FROM ( SELECT TOP 20
                                ROW_NUMBER() OVER ( ORDER BY rroad_sample_set.sample_set_number ASC ) AS scrolling_row_number,
                                rroad_sample_set.sample_set_number                                    AS sample_set_number
                           FROM rroad_sample_set
                          WHERE rroad_sample_set.sampling_id       = @sampling_id
                            AND rroad_sample_set.sample_set_number < @sample_set_number
                          ORDER BY rroad_sample_set.sample_set_number DESC ) AS older_sample_set
                 WHERE older_sample_set.scrolling_row_number = 20;

  23. MobiLink: Forgetting to call ml_add_column() when trying to use named parameters instead of "?" in versions 10 and 11 MobiLink scripts, resulting in a "What the ... ? Sassen Frassen Fricken Fracken!" moment during the first test (thank you, Jeff Albion).

  24. SQL: Omitting a PRIMARY KEY column from the WHERE clause, thus turning a singleton SELECT (or DELETE!) into something rather more enthusiastic than expected (thank you, Ron Hiner).

  25. HTTP web services: Leaving an & in the code when a ? is required, and vice versa, when editing service URLs; e.g., 'HTTP://localhost:12345/web_service&service_parm2=!parm2'

  26. SQL Anywhere: Forgetting that not all functions look like functions: SELECT CAST ( CURRENT TIMESTAMP, VARCHAR )

  27. Batch file: Trailing spaces on SET commands; e.g., SELECT CAST ( xp_getenv ( 'DEBUG_MESSAGES' ) AS VARCHAR ) returns 'OFF ' instead of 'OFF' after SET DEBUG_MESSAGES=OFF

  28. Forum: Clicking Reply on the main Question or Answer entry instead of the comment you wanted.

  29. SQL Anywhere: Forgetting to run dblog to tell the database file where the log is now, after moving the database and log files to a different folder (thank you, Justin Willey).

  30. SQL Anywhere: Having to look up WAIT in the Help ... every ... single ... time, to be reminded that's it's WAITFOR, not WAIT.

  31. SQL: Forgetting to check the SELECT against the GROUP BY, resulting in "Function or column reference to ... must also appear in a GROUP BY" (thank you, Glenn Paulley).

  32. SQL: Coding too much in the GROUP BY (like, say, the primary key) so every group contains but a single row (thank you, Glenn Paulley).

  33. Design: Forgetting to accomodate or prevent loops in a tree structure, resulting in a tree traversal process that pegs the CPU at 100%... forever (thank you, Ove B).

  34. MobiLink: Unwittingly using a variety of user ids when running sync*.sql, updating MobiLink scripts and running the MobiLink server, resulting in inexplicable inconsistencies.

  35. MobiLink: Accidentally creating multiple script versions and then getting them crossed up when updating MobiLink scripts and running the MobiLink client.

  36. SQL Anywhere: Forgetting to run the 32-bit version of SQL Anywhere when working with Excel proxy tables.

  37. ODBC Administrator: Running the 64-bit version (huh?) of odbcad32.exe (say what?) when you need 32-bit version at C:\WINDOWS\SysWOW64\odbcad32.exe (oh, fer #*@&!!!)

  38. ODBC Administrator: Forgetting to click OK ... twice ... to actually save your new ODBC DSN after celebrating your success with Test Connection.

  39. ODBC Administrator: Setting up an ODBC DSN on the wrong computer: "It goes with the client!" ... but sometimes it's not obvious where the client is actually located.

  40. Security: Forgetting which Windows user id you're using on which system, then spending too much time with Windows menus, firewall software and Google searches before the "Doh!" moment.

  41. SQL: Getting an exception that is not only completely inexplicable, but absolutely impossible for the statement that raised it... until you think to look inside the triggers.

  42. SQL Anywhere: Getting an exception because a FOR loop variable has a scope conflict with a column name, or worse, NOT getting an exception, just a wrong result.

  43. SQL: Forgetting the comma between two columns in SELECT list, thus turning the second column name into a profoundly misleading alias name for the first column.

  44. SQL Anywhere: Coding SET OPTION instead of SET TEMPORARY OPTION, then wondering where the commit came from.

  45. SQL: Not bothering to check that you have the same number of columns and arguments in an INSERT statement and then wondering why you have an error (thank you, Justin Willey).

  46. SQL: Forgetting to omit your autoincrement primary key from your INSERT column name list (thank you, Justin Willey).

  47. Foxhound: Changing the "Default" Monitor option settings, then expecting them to apply to an existing target database.

  48. Foxhound: Creating two separate Foxhound Monitor sessions for the same actual target database, perhaps one using a DSN and the other a DSN-less connection string.

  49. SQL Anywhere: Forgetting to code AUTOSTOP OFF on a START DATABASE statement, then wondering why the database immediately stopped after starting.