Re: Oracle 8.1.5 Memory Usage

Van:

Are you sure that trace files are written to RAM and not disk? I always thought the point of a trace file was to find fatal (and other) errors. You would think they would be flushed to disk immediately. Can't find an error if the task died with the trace still in core, can you? :-)

Anyway, all leads are good ones. I'll check it out.

   Eric Raskin

Van Messner <vmessner_at_netaxis.com> wrote in message news:dHGY3.22350$YI2.1017407_at_typ11.nn.bcandid.com...
> Hello Eric:
>
> These are very small databases, but even so block buffers is set too
> small. There are some Net8 trace files that can use all your memory very
> quickly if they are on. From where I'm sitting now I don't recall the
exact
> names or locations but it's one more thing to check.
>
> Van
>
> Eric Raskin <eraskin_at_paslists.com> wrote in message
> news:s36agmmkhsq94_at_corp.supernews.com...
> > I have a Solaris 7 Intel server running here with 512MB of RAM. I have
> two
> > databases at the moment: one is my "MAIN" database and holds my
corporate
> > data as well as an Oracle Designer repository. The other is a database
> for
> > one of my customers.
> >
> > The problem I'm having is that when I start up both databases, I eat up
> all
> > 512MB of RAM and then some!! I've been playing with settings in the
> > init*.ora files all day, but can't seem to shrink it down. I run out of
> > swap space and Oracle stops working. Once it even core-dumped. Here
are
> > the (I think) relavent settings from my init*.ora files:
> >
> > MAIN:
> > db_files = 80
> > db_file_multiblock_read_count = 16
> > db_block_buffers = 512
> > shared_pool_size = 10000000
> > java_pool_size = 5000000
> > processes = 50
> > log_buffer = 32768
> > db_block_size = 2048
> > mts_dispatchers = "(protocol=TCP)(DISP=1)(mul=ON)(pool=ON)"
> > mts_max_dispatchers = 2
> > mts_servers = 1
> > mts_max_servers = 5
> > compatible = "8.1.0"
> > open_cursors = 400
> > max_enabled_roles = 100
> >
> > Customer:
> > db_block_buffers = 512
> > shared_pool_size = 5000000
> > large_pool_size = 1000000
> > java_pool_size = 1000000
> > processes = 50
> > log_buffer = 32768
> > db_block_size = 2048
> > mts_dispatchers = "(protocol=TCP)(DISP=1)(mul=ON)(pool=ON)"
> > mts_max_dispatchers = 2
> > mts_servers = 1
> > mts_max_servers = 5
> > compatible = "8.1.0"
> >
> > Here is a "top" sorted by memory size:
> >
> > last pid: 19892; load averages: 0.02, 0.02, 0.06
> > 17:16:42
> > 83 processes: 80 sleeping, 2 zombie, 1 on cpu
> > CPU states: 99.4% idle, 0.1% user, 0.1% kernel, 0.4% iowait, 0.0%
swap
> > Memory: 512M real, 55M free, 578M swap in use, 108M swap free
> >
> > PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND
> > 19870 oracle 4 58 0 66M 32M sleep 0:00 0.00% oracle
> > 19854 oracle 14 58 0 62M 28M sleep 0:00 0.00% oracle
> > 19856 oracle 14 58 0 62M 28M sleep 0:00 0.00% oracle
> > 19858 oracle 14 58 0 62M 27M sleep 0:00 0.00% oracle
> > 19852 oracle 4 58 0 62M 27M sleep 0:00 0.00% oracle
> > 19877 daemon 4 58 0 61M 30M sleep 0:00 0.00% oracle
> > 19879 daemon 4 58 0 61M 29M sleep 0:00 0.00% oracle
> > 19862 oracle 4 58 0 61M 29M sleep 0:00 0.00% oracle
> > 19860 oracle 4 58 0 61M 29M sleep 0:02 0.00% oracle
> > 19868 oracle 4 58 0 60M 26M sleep 0:00 0.00% oracle
> > 19866 oracle 4 58 0 60M 26M sleep 0:00 0.00% oracle
> > 19835 oracle 4 58 0 58M 24M sleep 0:00 0.00% oracle
> > 19821 oracle 14 58 0 53M 19M sleep 0:00 0.00% oracle
> > 19823 oracle 14 58 0 53M 19M sleep 0:00 0.00% oracle
> > 19825 oracle 14 58 0 53M 19M sleep 0:00 0.00% oracle
> >
> > (sorry about the formatting...)
> >
> > Here is a top without ORACLE running:
> >
> > last pid: 19929; load averages: 0.03, 0.02, 0.05
> > 17:19:31
> > 58 processes: 55 sleeping, 2 zombie, 1 on cpu
> > CPU states: 99.8% idle, 0.1% user, 0.1% kernel, 0.0% iowait, 0.0%
swap
> > Memory: 512M real, 153M free, 84M swap in use, 603M swap free
> >
> > PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND
> > 19929 root 1 51 0 1796K 1168K cpu1 0:00 0.24% top
> > 19738 eraskin 1 58 0 1820K 1132K sleep 0:03 0.05% top
> > 19894 root 1 48 0 2012K 1420K sleep 0:00 0.05% bash
> > 1 root 1 58 0 624K 224K sleep 0:09 0.04% init
> > 7450 root 1 58 0 2000K 1140K sleep 0:05 0.01% in.named
> > 319 root 8 58 0 9556K 7792K sleep 0:42 0.00% jre
> > 464 daemon 4 58 0 19M 11M sleep 2:16 0.00% dwhttpd
> > 657 root 12 60 0 1864K 1348K sleep 1:42 0.00% mibiisa
> > 389 root 8 58 0 9416K 7512K sleep 0:36 0.00% jre
> > 573 root 14 58 0 7136K 3988K sleep 0:32 0.00% httpd
> > 601 root 8 58 0 11M 9268K sleep 0:19 0.00% java
> > 440 root 11 58 0 7424K 3244K sleep 0:15 0.00% httpd
> > 409 root 1 58 0 1736K 1068K sleep 0:04 0.00% nmbd
> > 259 root 15 54 0 2948K 1320K sleep 0:02 0.00% syslogd
> > 274 root 10 15 0 2072K 1420K sleep 0:01 0.00% nscd
> >
> > As you can see, the almost all of my swap area gets used before anybody
> even
> > connects to the databases. What am I doing wrong? Do I just need more
> RAM?
> >
> > TIA
> >
> > --
> > Eric Raskin eraskin_at_paslists.com
> > Prof. Advertising Systems Inc. 914-741-1100
> > 70 Memorial Plaza Fax: 914-741-2788
> > Pleasantville NY 10570
> >
> >
>
>
Received on Wed Nov 17 1999 - 19:10:48 CST