Re: Backend memory dump analysis

It looks much better.

>While I didn't do anything about it here, I think it'd likely be a
>good idea for MemoryContextStats printout to truncate the context ID
>strings at 100 characters or so

It would be great if there was an option to show full sql.
For instance, current statistics is not sorted (should it be?), so it just
prints the first 100 items no matter the size of entries.
Luckily there's a function that accepts "number of nodes to print" as an
argument.
It would be better if it printed the 100 top consumers (including
descendants) though.

I think it makes sense to move all the numerics to the front, so the
numbers are located in the more or less the same columns.
Current output is hard to read as you basically have to search for "free"
and "used" markers.

What do you think of (number of bytes are printed with 6 characters, and
number of chunks with 2 characters)

  16384 total in  2 blocks;  12448 used;   3936 free ( 5 chunks):
PL/pgSQL function tg_pslot_biu()

instead of

PL/pgSQL function tg_pslot_biu(): 16384 total in 2 blocks; 3936 free
(5 chunks); 12448 used

?
I think "used memory" is more important than "free". As far as I
understand, the main use-case is "analyze memory consumption", so one cares
"what is consuming the memory" more than "what context have enough free
space".

PS. "TopMemoryContext: 2143904 total" and "Grand total: 13115912 bytes"
does confuse. It basically says "TopMemoryContext is 2MiB, and grant total
is somehow 12MiB". It does not clarify if totals are "self memory" or
"including descendant contexts". In your case the difference is 10MiB, and
it is not that obvious why is that.

Vladimir