Re: Is my database now too big?

Scott Marlowe wrote:
> ...
>
> Again, I'm kinda shooting in the dark here as you reveal more and more
> what you are doing a little at a time.  A test case that can invoke
> this failure would be most useful.
>
After seeing this today:
ERROR:  duplicate key violates unique constraint "ers_pkey"
ERROR:  duplicate key violates unique constraint "foo_pkey"
ERROR:  duplicate key violates unique constraint "foo_pkey"
ERROR:  duplicate key violates unique constraint "foo_pkey"
ERROR:  duplicate key violates unique constraint "foo_pkey"
ERROR:  duplicate key violates unique constraint "foo_pkey"
ERROR:  could not open segment 1 of relation 1663/10793/2659 (target
block 858862642): No such file or directory
ERROR:  could not open segment 1 of relation 1663/10793/2659 (target
block 858862642): No such file or directory
ERROR:  could not open segment 1 of relation 1663/10793/2659 (target
block 858862642): No such file or directory
ERROR:  could not open segment 1 of relation 1663/10793/2659 (target
block 858862642): No such file or directory
...

...there was little or no activity during this time, apart from
some inserts, maybe some selects, etc.  Nothing that should
have caused this kind of upset.

There is a file that matches this:
-rw-------  1 postgres  wheel  57344 Oct 14 22:57
/data/db/second/base/10793/2659
but it isn't in the directory where I moved most of the indexes to:
ls /data/index/ext/10793/
16390  16397  16399  16406  16407  16410  16414  16425  16434  16435

I don't know if the file numbers have any meaning?

But in addition, the list of tables (\dp) is now fubar'd.

I'm starting to wonder if it is a combination of:
- the operating system (NetBSD 4.99.20)
- the hardware (small HP box, not meant for hard work like this but
shouldn't be impossible for it)
- the way pkgsrc compiles postgresql for NetBSD

I'm shying away from the hardware (or at least RAM/CPU) because I'd
expect there to be some other kind kinds of faults show up, ultimately
leading to a panic due to just random corruption of some kernel data
structure.  As it is, everything else seems to be functioning ok.

Darren