Re: Select Query hangs up on IDS 9.40.
Gary Quiring reported queries that hang on IDS 9.4 but work on 7.31 with OPTCOMPIND 0, even with the latest dostats on a 9.4UC5 dev server. Art Kagel asked how the 7.31 query plan differed from the 9.40 one. The thread ends there with no resolution.
Auto-generated by Claude from the posts below — may be imperfect; read the full thread.
> Which definitely points to a data distribution problem. Just that fact that > the cost of the selected query plan is an order of magnitude higher than the > ones produced changing something is a clue. Are you certain that the stats > were recalculated AFTER the copy to 9.40? How were stats produced? What > release of 9.40? > > Art S. Kagel Hi Art, Vipul works as the programmer here with me. We use your dostats and run it daily with these options: dostats -b -B 3. The IDS is 9.4UC7. We are still using the old dostats 4.12 source 1.01. I did get the new copy and it is running on the dev server but we have not moved it to production. Would that be an issue? Thanks Gary Quiring
Gary Quiring wrote: >>Which definitely points to a data distribution problem. Just that fact that >>the cost of the selected query plan is an order of magnitude higher than the >>ones produced changing something is a clue. Are you certain that the stats >>were recalculated AFTER the copy to 9.40? How were stats produced? What >>release of 9.40? >> >>Art S. Kagel > > > Hi Art, > > Vipul works as the programmer here with me. We use your dostats and > run it daily with these options: dostats -b -B 3. The IDS is 9.4UC7. > We are still using the old dostats 4.12 source 1.01. I did get the new > copy and it is running on the dev server but we have not moved it to > production. Would that be an issue? It _might_ be. Probably not, but running the new one is worth a shot. Art S. Kagel
> > It _might_ be. Probably not, but running the new one is worth a shot. > > Art S. Kagel We can reproduce the problem on our DEV server which is 9.4UC5. I ran with the latest dostats (it works no more core dump) and it did not resolve the problem. I just want to make clear these same queries with the same schema and data work fine on 7.31 with OPTCOMPIND = 0. Thanks Gary Quiring
Gary Quiring wrote: >>It _might_ be. Probably not, but running the new one is worth a shot. >> >>Art S. Kagel > > > We can reproduce the problem on our DEV server which is 9.4UC5. I ran > with the latest dostats (it works no more core dump) and it did not > resolve the problem. > > I just want to make clear these same queries with the same schema and > data work fine on 7.31 with OPTCOMPIND = 0. > > Thanks > Gary Quiring > How does the query plan that 7.31 is producing differ from the one 9.40 is using if at all? Art S. Kagel