Re: Help with count(*)

After a long battle with technology, hannu@tm.ee (Hannu Krosing), an earthling, wrote:
> Christopher Browne kirjutas R, 14.11.2003 kell 16:13:
>> Martha Stewart called it a Good Thing when mallah@trade-india.com (Rajesh Kumar Mallah) wrote:
>> > INFO:  "profiles": found 0 removable, 369195 nonremovable row versions in 43423 pages
>> > DETAIL:  246130 dead row versions cannot be removed yet.
>> > Nonremovable row versions range from 136 to 2036 bytes long.
>>
>> It seems as though you have a transaction open that is holding onto a
>> whole lot of old rows.
>>
>> I have seen this happen somewhat-invisibly when a JDBC connection
>> manager opens transactions for each connection, and then no processing
>> happens to use those connections for a long time.  The open
>> transactions prevent vacuums from doing any good...
>
> Can't the backend be made to delay the "real" start of transaction until
> the first query gets executed ?

One would hope so.  Some time when I have the Round Tuits, I ought to
take a browse of the connection pool code to notice if there's
anything to notice.

The thing that I keep imagining would be a slick idea would be to have
a thread periodically go through once for however many connections the
pool permits and fire a short transaction through every
otherwise-unoccupied connection in the pool, in effect, doing a sort
of "vacuum" of the connections.  I don't get very favorable reactions
when I suggest that, though...
--
(reverse (concatenate 'string "ac.notelrac.teneerf" "@" "454aa"))
http://cbbrowne.com/info/sgml.html
Rules  of  the  Evil Overlord  #80.  "If  my  weakest troops  fail  to
eliminate a  hero, I will send  out my best troops  instead of wasting
time with progressively stronger ones  as he gets closer and closer to
my fortress." <http://www.eviloverlord.com/>