What does
nofiles(descriptors) 2000
in the ulimit -a output stand for?
Is this the one that needs to be increased when the db cfg parm MAXFILOP is increased?
I must not have read it correctly - and I wasn't aware that it was possible. Thank you for the information - I like it better than describe since it actually gives you the column length in bytes.
A select(*) is not the same as describe. Besides, if it is a large table you will be accessing disk unnecessarily.
Can you go against the system catalog? Then you can try accessing syscat.columns and syscat.keycoluse giving it the tabschema and tabname.
That is strange. If you look at the previous post of yours on loading a large table, you should be doing this in the same statement.
Look at
http://www.tek-tips.com/gviewthread.../178/qid/725298
However, a word of caution - if you are in a recovery situation, and are rolling forward through...
Then I think you can autoload replace the data keeping indexes and PK on. I think it amounts to the same thing since either you build the indexes or autoload takes care of it in the build phase.
The reason you cluster data is for read performance. Are you always going to replace data? How do you intend to load data - with autoloads or inserts? If you choose autoloads, then remember that autoloads do not look for freespace - they start an extent with each run and load the data in. And if...
Actually, I think the not logged initially is not in effect in this case. I have found that they have to be issued in the same statement or it does not work(ie,logs).
I have found this to work. Run it is a shell script. Hope this helps.
. /udbhome/<instname>/sqllib/db2profile...
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.