Jyotika,
Hmm, this doesn't look like it's going to be easy to
track down. If you haven't already done so, I'd
recommend logging a tar with support on this.
A few observations:
1. By default MTS is configured on 8i so make sure
you aren't unwittingly using it by checking the
init.ora and/or querying V$PARAMETER for MTS_
parameters. When MTS is configured, connections will
default to shared connections unless the connect
string specifically states to make it a dedicated
connection.
2. Whether the connection is dedicated or MTS is
determined by the connect string used in the USING
clause when creating the db link. The dblink you've
referenced is the private db link owned by REPADMIN.
So you should have a public db link of the same name
that specifies the connect string (HOST in
DBA_DB_LINKS). Check that entry in the tnsnames.ora
to see if it includes a SERVER=DEDICATED entry (see
Note: 73145.1). This is relevant only if you've
determined that MTS *is* being used.
3. How frequent are these errors? When did they
start happening? What's changed since they started
happening? Is there any pattern to when the errors
occur (e.g. heavy load, network latency, etc...).
4. Do you have a middle tier involved? If so, it
could be that the middle tier's session is timing out
to the db. Are there any timeout or profiles set for
the sessions in the db or on the middle tier?
5. You mention that it happens with transactions
against both the synchronous and asynchronous, but
that doesn't sound right, unless you're using the
serial push mechanism (PARALLELISM=>0) in
dbms_defer_sys.push. If it were the parallel push
getting this error this would be recorded in the SNP
trace file when the job failed to push. The fact that
this is being returned to the user session sounds like
it's the synchronous replication having the problem.
Can you try setting all groups to asynch using
parallel push (use
dbms_repcat.alter_master_propagation to change the
propagation mode for a group; the group must be
quiesced first).
If none of the above suggestions help you're most
likely going to need Net8 tracing to nail this down
and that'll be messy since it affects ALL connections
and you're going to need client and server side
tracing for the entire transaction.
HTH,
-- Anita
--- Jyotika Bhandarkar <[Email Address Removed] wrote:
> Hello Anita,
>
> Thanx for your reply..
> The Oracle version I am using is 8.1.7.0.1
> The application (Multithreaded application) is
> written in Java so it uses JDBC to
> make connections with the database (which is not
> configured to use MTS (multi
> threaded Server)).
> I have three replication groups (2 Synchronous &
> 1 asynchronous) and 2
> master(replication) sites. I have created one
> database link (on each of the
> replication site), which is used for replication as
> follows:
> create database link myLink connect to
> repadmin identified by repadmin;
> This link is created during the configuration of
> replication.
> How do I use a dedicated server connection in
> creating a database link ?
> Moreover, I have seen this problem for
> transactions that deal with both
> asynchronous & synchronous tables.
>
> Thanx & Regards,
> Jyotika
>
> "A. Bardeen" wrote:
>
> > Jyotika,
> >
> > Soulds like part of your transaction is timing out
> > (e.g. profile limit exceeded, etc...). Are you
> using
> > MTS connections?
> >
> > What version of Oracle? Are the transactions
> being
> > pushed synchronously or asynchronously? I
> recommend
> > using a dedicated server connection for the
> connect
> > string used by the dblinks involved in
> replication.
> >
> > HTH,
> >
> > -- Anita
> >
> > --- Jyotika Bhandarkar <[Email Address Removed] wrote:
> > > Hi,
> > >
> > > My application keeps encountering oracle
> errorcode
> > > - ORA-02055. This
> > > is invariably followed by ORA-02067.
> > > Just an FYI - The application runs against a
> > > database which is
> > > replicated (using Oracle8i's Multi-Master
> > > replication). It makes around
> > > 40 oracle connections at startup. So it is
> possible
> > > that a number of
> > > these connections could be catering to different
> > > distributed
> > > transactions at any given point of time.
> > >
> > > What causes oracle to give this error code
> > > (ORA-02055) ? Could it be
> > > a oracle configuration issue ?
> > >
> > > Can anybody help ?
> > >
> > > Thanks in advance
> > > Jyotika
__________________________________________________
Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger
http://phonecard.yahoo.com/
Oracle LazyDBA home page