Обсуждение: Fwd: Request for error explaination || Adding a new integer in indextupleData Structure

Rohit Goyal <rhtgyl.87@gmail.com> writes:
> Hi All,
> I was trying to modify indextupledata structure by adding an integer
> variable. ButI faced an error message "psql: FATAL:  could not find tuple
> for opclass 10032".

> Could anyone please help me in resolving this issue.

You broke a system catalog index.  Without seeing what you changed and
where, it's impossible to say just how, but that's the bottom line.

This is the first line which i want to write in the code. Can you tel me how to add an integer in  indextupledata structure if I have a fresh system with no changes?

In recent versions of PG, opclass 10032 is btree name_ops (unless you've
also added/removed system catalog entries), which is a pretty plausible
thing to be one of the first indexscanned fetches during relcache.c
initialization, so I don't think there's any great significance in this
particular error message.  It's likely that you broke *all* indexscans
not just one specific one.

I am not sure how i can break all indexscan as this is the first line I wrote in the code.  

 

                        regards, tom lane

Hello,

I started all the process again and configured my eclipse with raw postgresql code. First change i made in the code is

I added int i; in indextupleData structure in itup.h.

I got the same error message. Please help me to understand and solve the issue. I want to add an integer in index tuple for btree.

Hello,

I started all the process again and configured my eclipse with raw postgresql code. First change i made in the code is

I added int i; in indextupleData structure in itup.h.

You should show us *exactly* where you added it.  (Doing so is what "diff" was developed for, so please use that or a similar tool.)

 

I got the same error message. Please help me to understand and solve the issue. I want to add an integer in index tuple for btree.

The data from IndexTupleData is written to disk, and then read back in again.  Did you initdb a new database cluster after you made your change?  If you did the initdb with the original code, and then tried to point your new code at the old disk files, that is very unlikely to work, as format is now different.

Cheers,

Jeff