memory is not linear to work done - 2 transactions with 200k keys
will leak less than 1 transaction with 400k keys
Regarding Tom's question:
The old_prefs does not fit in work_mem but is quite small regarding the total
RAM. Isn't the "work_mem" a limit of the memory each backend could allocate for
sorting, grouping and aggregation? My understanding is that bigger allocation
will overflow to disk and will not kill the server. I could be wrong though.
Thanks in advance and best regards
--
Luben Karavelov
--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs
e finishes with 900M RSS.
After "ANALYZE old_prefs" the planner chooses GroupAggregate instead of
HashAggregate - you were right about missing statistics of old_prefs.
Thank you for figuring out this case
Best regards
--
luben karavelov
--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs
