Re: FW: SQL code -246, ISAM code -154 problem.

On Monday, January 20, 1997 2:57 PM, Peter = Jennings[SMTP:peter_jennings*@uk.ibm.com] wrote: [...We find that if two users try to update different rows in the same = child table, the first will suceed and the other fail with SQL code = -246, ISAM code -154.] I have read all the posts about this problem, and I found that Rob = Weight's one is quite helpfulI (4 me). I want to add some more here as = I am currently suffering with similar errors (ISAM -154 & -143), and I = was asking help from anywhere I can. Our application causes at least 70 = deadlocks a day! There are couple of articles are worthwhile to read: 1. IFX-FAQ Section 6.4: What error msgs are caused by locking problems 2. IFX-FAQ Section 6.7: Locking in pre-6.0 engines I'm on an Informix 5.05 UC3 engine, with buffered transaction logging & = low-level locking. If anyone knows how to (except upgrade to...) cope = with this problem, please let me know, thank you. There are serveral things we might try: 1. Change low-level locking to page lock or even table lock if = concurrent transaction number is not high Still low-level locking: 2. use dirty read or set wait time a bit longer 3. To avoid the 'scan-bug', try to select or join the indexed fields = only, or build tmp index(es). I think if it is a large scan, the engine = will actually build a tmp index for you 4. Make your indexes to unique as possible as you can to avoid the = 'Twin Problem', or simply just drop some of the indexes I have posted a question last week asking how to measure (a) table io on = informix 5.x; (b) timing a transaction & how much log space it has = used.......basically just try to find out where the contention is, = unfortunately, I haven't got a good answer yet. =20 Here is one more: how can I tell how much time a checkpoint has taken on = Informix 5.05? Thanks 4 ur time SEAN (swong@kato.den.csci.csc.com) Computer Science Corp