Question:
Hi
Some questions have been bugging me about process chains
Let me know if you have one or more answers.
Thanks,
C
1/ How can you merge two process chains? If you have PC1 and PC2 how do you make a PC that is equivalent to PC1+PC2 (without having to reenter from scratch the components of PC2 into PC1)?
2/ There is no documentation of process chains contents e.g. process chains are not in metadata repository. Do you know of another way to see what is in process chains (other than looking at RSPC)? e.g. table
3/ Do you know of a way to get a list of infopackages that are not in process chains (other than doubleclicking one infopackage at a time)? Is there a table that provides that info?
4/ Should all links be on ALWAYS rather than successful e.g. drop indexes then load infopackages whether or not successful. If not, if the drop index fails then the load will not go through and neither will be subsequent process chains?
5/ Is there a way to do a global find and replace in the description of infopackage? this is to keep consistency and naming conventions for packs created by several users? right now I am renaming one infopack at a time?
Answer:
How can you merge two process chains? If you have PC1 and PC2 how do you make a PC that is equivalent to PC1+PC2 (without having to reenter from scratch the components of PC2 into PC1)?
Ans: You cannot do that.But the work around is declare these 2 PCs as meta chain and call them in main PCs
Should all links be on ALWAYS rather than successful e.g. drop indexes then load infopackages whether or not successful. If not, if the drop index fails then the load will not go through and neither will be subsequent process chains?
No ,it should not be always... we create the index to improve data access performance to cube...if we dont drop the same before loading data then new data wont be properly indexed...so it should be always successful
Answer:
ans 3. table RSPCCHAIN gives all IP that are IN a PC. So all the rest are NOT.
Gr,
M.
Answer:
Thank you guys for sharing. That is what I thought. I just wish that process chains were little more flexible e.g. cut and paste, documentation etc Next release may be?
Need clarification on using "successful" when dropping indexes then load infopackages. In understand the new data wont be properly indexed...But what about subsequent loads in the chain? In case of failure of indexes drop/reload will the whole chain be cancelled?
As an example say I have a Chain including loads to two cubes (in seria):
drop cube 1 then
if successful load cube 1 then
if successful recreate cube 1 then
Always drop cube 2 then
if successful load cube 2 then
if successful recreate cube 2.
What happens if the drop cube 1 fails, will he whole chain cancel? Or will it go on to dropping cube 2?
Thanks,
Kathy
Answer:
In case of failure,PC will show red for dropping index on cube in the errorlog and will stop there itself.What you need to do is check why it has failed and correct the same.If you right click on the red block it will show you option of repeat.
Answer:
As an example say I have a Chain including loads to two cubes (in seria):
drop cube 1 then
if successful load cube 1 then
if successful recreate cube 1 then
Always drop cube 2 then
if successful load cube 2 then
if successful recreate cube 2.
What happens if the drop cube 1 fails, will he whole chain cancel? Or will it go on to dropping cube 2?
Ans : if the drop cube 1 fails the process chain wont get cancelled but it will show red in errorlog for this step.What you need to do is correct the error and then right click on error block say repeat.
Answer:
One more ?
I have a "transaction" process chain with 5 infopackages/cubes being loaded (in series with an "ALWAYS" link).
Then say Step 2 (load to cube 2) fails. What I do not understand right now is why the chain stops at step 2 and does not proceed with step 3 (ALWAYS)?
Thanks,
C