Re: performance problem - 10.000 databases

W liście z śro, 05-11-2003, godz. 19:23, Jeff pisze:
> On 05 Nov 2003 19:01:38 +0100
> Marek Florianczyk <franki@tpi.pl> wrote:
>
> > and it works better, but no revelation, when I do "\d
> > schemaname.table" it's better. I've to still wait about 10-30 sec. and
> > now it's only 100 clients connected. :(
> >
>
> So it only goes slow with hundred(s) of clients connecting and
> disconnecting?
>
> I'm guessing the CPU is pretty pegged on the box.


Thats not exacly what you think, my test script ( some post earlier ) is
doing queries, look at the results below, queries are very fast, just
these with "\d" ale slow:

test2364: connect:0 q_fast:0 q_med:0 q_slow:15 q_upd:0 q_upd_all:0
test2346: connect:0 q_fast:0 q_med:0 q_slow:17 q_upd:0 q_upd_all:7
test2347: connect:0 q_fast:0 q_med:0 q_slow:15 q_upd:0 q_upd_all:9
test2350: connect:0 q_fast:0 q_med:0 q_slow:16 q_upd:0 q_upd_all:8
test2349: connect:0 q_fast:1 q_med:0 q_slow:15 q_upd:0 q_upd_all:8
test2343: connect:0 q_fast:1 q_med:0 q_slow:17 q_upd:0 q_upd_all:7
test2351: connect:0 q_fast:0 q_med:0 q_slow:17 q_upd:0 q_upd_all:9

output from command: top

 02:20:00 up 1 day,  6:19,  2 users,  load average: 6.62, 6.67, 6.60
130 processes: 123 sleeping, 7 running, 0 zombie, 0 stopped
CPU states:  82.4% user,  16.2% system,   0.0% nice,   1.4% idle
Mem:   2070084K total,  2052932K used,    17152K free,    78048K buffers
Swap:   995988K total,     6272K used,   989716K free,  1771140K cached

  PID USER     PRI  NI  SIZE  RSS SHARE STAT %CPU %MEM   TIME COMMAND
18775 postgres  18   0  5532 5440  4688 S     8.7  0.2   0:01 postmaster
19001 postgres  15   0  5484 5392  4664 S     8.3  0.2   0:00 postmaster
19085 postgres  13   0  5492 5400  4680 S     8.3  0.2   0:00 postmaster
18831 postgres  16   0  5532 5440  4680 S     7.9  0.2   0:01 postmaster
19114 postgres  19   0  5436 5344  4624 S     7.9  0.2   0:00 postmaster