![](//teradataforum.com/imgs/ptdfw1.gif) |
![](//teradataforum.com/imgs/ptdfw2.gif) |
Archives of the TeradataForum
Message Posted: Mon, 01 Apr 2002 @ 16:45:22 GMT
Subj: | | Re: Question on atomicity in CLI |
|
From: | | James A. Judge |
Assuming the session does not logoff during steps 1-4, S1 update will complete but the transaction will not be committed until the "et;"
is received in S3. If a restart or session disconnect occurs after "update" in S1 but before S3, then the S1 "update" is rolled back. If S3
then processes, and the S3 "update" is successful the S3 "update" is committed because the previous "bt;" is no longer in effect; and the S3
"et;" will get an error message, but not roll back the S3 "update". So you could have an incorrect "unit-of work" (S1 & S3 updates both
complete) due to roll-back between S1 and S3.
I haven't worked with CLI in awhile and this is Teradata mode of transaction processing due to BT; & ET; but you need to have code to
check for restart and error messages during S2 that would allow you to reprocess S1. Appendix B of CLIv2 manual talks to this, I
think.
| |