Connection Blues, Solaris

From: Lane W. Sharman <lshaman1_at_san.rr.com>
Date: 1997/05/02
Message-ID: <336ABB95.6468@san.rr.com>#1/1

Hi Everyone,

Well, here is my tale of woe. ANY and ALL input is appreciated in advance. I manage a fairly large online web community, flirt.com, and have transferred it to a netra i-140. The problem is that I cannot get listener.ora to run any pre-spawned threads without hanging connections. With pre-spawning and MTS, the multi-threaded server, connections ought to be like lightening. Well, they are fast when they work but they hang every couple of connections. This would be the ideal set up according to the docs: a dispatcher pool within the instance and the listener pre-positioning connection threads per connection request. Is there a patch/upgrade to make this work? Environment:

Platform=netra i/140
OS=Solaris 2.5.1
Oracle=7.3.2, Workgroup Server
Application Library=DBI/DBD running on server Memory=192MB

When I remove the threads, of course, the system works properly; the connections become reliable, but simultaneous requests get serialized leading to unacceptable wait times.

Also, is there a memory leak, severe or mild, in this version of the OCI? Here is listener.ora (some values etched out to protect the innocent):

LISTENER =
  (ADDRESS_LIST =

        (ADDRESS = 


          (PROTOCOL = TCP)

          (Host = xx.yyy.zz.qqq)

          (Port = 1521)

          (QUEUESIZE=80)

        )

  )

STARTUP_WAIT_TIME_LISTENER = 0
CONNECT_TIMEOUT_LISTENER = 100
TRACE_LEVEL_LISTENER = OFF
SID_LIST_LISTENER =
  (SID_LIST =
    (SID_DESC =

      (SID_NAME = rrrr)
      (ORACLE_HOME = /opt/oracle7)
			(PRESPAWN_MAX = 80)
			(PRESPAWN_LIST=
				(PRESPAWN_DESC=(PROTOCOL=tcp) (POOL_SIZE=20) (TIMEOUT=10))
			)

    )
  )

-- 
Lane Sharman
Pour chasser le chagrin,
il faut du vin.

Received on Fri May 02 1997 - 00:00:00 CDT