[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference orarep::nomahs::dec_data_distributor

Title:The Replication Option for Rdb
Notice:Product renamed to Replication Option for Rdb
Moderator:BROKE::PROTEAU
Created:Wed Mar 02 1994
Last Modified:Wed Jun 04 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:287
Total number of notes:1231

283.0. "DDAL-E-SQLCODERR severe transfer failure" by CHSR38::RROHR (Cajun? Zeydeco? Both!) Tue Apr 29 1997 10:34

    I have a brand new installation of ROR V7 with Rdb 7.0 ECO1 on an AXP
    62. I can't get an extraction to run. There are no tables created in
    the source db or in the target db, but the transfer is marked with 
    status suspended in the transfer db.
    
    It fails with an error accessing the transfer db. Though that's the
    only db it actually writes to... Is there any debug flags or such stuff
    to get some insight into this problem? I read other notes with this
    problem, and I have restarted the monitor, recreated the transfer
    database, changed from 'new' to 'existing' target, restricted the
    transfer to 1 table only, and am out of ideas now. The db's are all
    mine, and I am running with defpriv=all, so I can't imagine a priv problem.
    
    thanks for any hint/pointer.
    /Regina
    
    The error message is:
    
    copy process spawned for transfer ROR_TEST                       
    scheduled to be executed at  29-APR-1997 15:12:09.18
    there are  1 copy processes active
    
    ....
    
    Copy process completion acknowledgement received
    
    
    
    ----- 29-APR-1997 15:12:15.18 ----- TRANSFERS      
    -------------------------
    
    execution of transfer ROR_TEST                       
    finished at  29-APR-1997 15:12:15.02
    
    severe transfer failure; not retrying
    there are  0 copy processes active
    
    and from the target logfile:
    
    Transfer database = 38_W2:[ROHR]DDAL$TR_DB
    Transfer name     = ROR_TEST                       
    Transfer option   = E (Extraction transfer)
    
    15:12:11  %DDAL-I-READTRDEF, reading the transfer definition from the
    transfer d
    atabase
    ----- 29-APR-1997 15:12:14.53 ----- Error          
    -------------------------
    %DDAL-E-SQLCODERR, status code   -1   received, indicating an error in
    accessing transfer database
    
                                   
T.RTitleUserPersonal
Name
DateLines
283.1Some things to investigateBROKE::PROTEAUJean-Claude ProteauTue Apr 29 1997 12:2811
Here's a reply I got from Shou-pin Yu:

Without an in-depth look at the problem , I would suggest to check a few
things:

1. the definition of the system logical name for the transfer database
2. whether the user can attach the transfer database using the logical
   name and do a SELECT on ROR tables.
3. if both are fine, then turn on dispatch debug flag so see if we can get
   more info as to what has transpired in attaching or accessing TR DB.

283.2More infoCHSR38::RROHRCajun? Zeydeco? Both!Tue Apr 29 1997 12:5921
(LNM$SYSTEM_TABLE)

  "DDAL$TRANSFER_DATABASE" = "38_W2:[ROHR]DDAL$TR_DB"
  "DDAL_TRANSFER_DATABASE" = "38_W2:[ROHR]DDAL$TR_DB"

SQL> att 'f ddal$transfer_database';
SQL> select * from ddal_tnsf_status; 
 TRANSFER_NAME                      REINI_FLAG   TRANSFER_STATE   
   LAST_COMPL_STATUS   LAST_TRANSFER_TIME        STATE_CHANGE_TIME         
 ROR_TEST                                    0                4   
            19301275   17-NOV-1858 00:00:00.00   29-APR-1997 15:12:15.17   
 
1 row selected
    
    I have noted down that the DSP_DEBUG_FLAGS logical is defined in
    rdb$client(server)_defaults.dat. But those are only for remote access,
    I thought? Is this the correct place to define the dispatch debug?
    
    Thanks,
    Regina
                     
283.3Try thisBROKE::PROTEAUJean-Claude ProteauTue Apr 29 1997 13:2530
    
    Regina,
    
    I'm not certain where the debug flags should be set (I personally
    haven't had to do it often, so I forget).  Let's forget that idea for
    the moment and come back to it later if necessary.
    
    Remember that when the transfer is executed, the copy process is
    started up in a detached process running under your login account (you
    being the one who created the transfer definition).  Particularly in a
    multiversion Rdb environment, it is easy to forget to set things up
    correctly for a detached process.
    
    I suggest you try the following.  Modify your transfer definition to
    point to a prologue file (A DCL COM file of your own making).  For
    example,
    
    	ALTER TRANSFER transfer_name PROLOGUE FILE IS filename;
    
    Create a DCL COM file to invoke SQL.  Ask SQL to SHOW VERSION.  Then try
    to attach to the transfer database and do a select * from
    ddal_tnsf_status again.  I suspect the prologue will get an error,
    indicating that your environment is not properly set up for transfer
    execution.
    
    If this turns out to be the problem, take another look at the ROR 7.0
    handbook.  If you cannot find an explanation (something you missed),
    let me know.
    
    Claude
283.4Shame on me...CHSR38::RROHRCajun? Zeydeco? Both!Wed Apr 30 1997 05:408
    Ahem, well it helps when you have the documentation on paper and not
    only on screen...
    
    I take the blame, but I guess DDAL could pass back the Rdb error
    message (which is pretty explicit in such cases).
    
    Thanks a bunch,
    Regina                                         
283.5BROKE::YUWed Apr 30 1997 22:3914
Regina,

In ROR V7.0, all data accesses to the Transfer Database are made through
the SQL interface, rather than the DSRI calls.  One impact of this change
is that we have no direct access to the Rdb message vector.  We can only
check SQLCODE to see if there is an error and to report its value.  If the
value is a negative number other than '-1', we can look up the table in
SQL manual to understand the nature of the error.  The value '-1' is a
'catch-all' SQL error code.  In this case, we need to look into the
circumstances to further determine the cause of an error.

Regards,

Shou-pin
283.6NOVA::SMITHIDon't understate or underestimate Rdb!Thu May 01 1997 14:354
You do have access to the message vector and you can get the message for the
log.

Ian