[Search for users]
[Overall Top Noters]
[List of all Conferences]
[Download this site]
Title: | Reliable Transaction Router |
|
Moderator: | TALER::DESHMUKH |
|
Created: | Tue Dec 12 1989 |
Last Modified: | Thu Jun 05 1997 |
Last Successful Update: | Fri Jun 06 1997 |
Number of topics: | 695 |
Total number of notes: | 2564 |
691.0. "count the timeout value as server's processing time" by DEKVC::JONGHOLEE () Mon May 12 1997 11:18
Hi,
My customer (KSE, Korea Stock Exchange) has some questions about declaring
RTR transaction timeout value and ENQing message.
(1) Question on declaring timeout value
Client Program Server Program
-------------- --------------
SYS$START_TXW SYS$DEQ_TXW
with 20 sec. timeout if rtr$_starttx
SYS$ENQ_TXW SQL processing
SYS$DEQ_TXW if rtr$_votetx
SYS$COMMIT_TXW SYS$VOTE_TXW
SQL COMMIT
Please inform me whether there are any way to proceed to next step of
"SYS$ENQ_TXW" of client program, after server program receives
the message by "SYS$DEQ_TXW" ?
That is, serveral ENQed messages from clients will be waited on
transaction router to receive one server's service. Also, declared
RTR transaction timeout value is counted from "SYS$START_TX" until
"SYS$COMMIT_TXW".
But, my customer wants to count the timeout value as server's
processing time of client's message.
(2) Please advise me on my customer's program structure.
Client Program Server Program
-------------- --------------
SYS$START_TX SYS$DEQ_TXW
SYS$ENQ_TXW if rtr$_starttx or rtr$_normal
SYS$ENQ_TXW SQL processing
SYS$ENQ_TXW if rtr$_votetx
SYS$ENQ_TXW SYS$VOTE_TXW
SYS$ENQ_TXW SQL COMMIT
SYS$COMMIT_TX
(2-1) If client program ENQs messages continuously, but DEQ cycle of
server program was delayed very long time for some reasons,
what expected problems are ?
(2-2) Is it possible client program ENQs next message in
a RTR transaction, after server program DEQed current one from
client ?
If you need more descriptions about my questions, please inform me.
Best Regards,
Jong-Ho Lee.
T.R | Title | User | Personal Name | Date | Lines |
---|
691.1 | | AMCFAC::RABAHY | dtn 471-5160, outside 1-810-347-5160 | Mon May 12 1997 18:36 | 22 |
| re .0:
For the client program to know that a server has begun working on its'
transaction, I suggest the following construct;
client server
------ ------
$dcl_tx_prc/req $dcl_tx_prc/server
$start_tx
$enq_tx $deq_tx
$deq_tx $enq_tx
DB processing
$setimr/daytim=20
$deq_tx $enq_tx
$wflor
if timer expired
then $abort_tx
...
The first reply is sent be the server program before doing any DB work as a
single to the client program. The second reply contains the results of the
query.
|
691.2 | | AMCFAC::RABAHY | dtn 471-5160, outside 1-810-347-5160 | Mon May 12 1997 18:38 | 5 |
| re .0:
There are flow control mechanisms which will inhibit the client program from
continuing to function. Specifically, the process will end up in the RWAST
state until the server program catches up some.
|
691.3 | Thanks for you | DEKVC::JONGHOLEE | | Mon May 19 1997 10:40 | 6 |
|
Thanks for your advice.
Regards.
Jong-ho Lee.
|