ORA-03113: end-of-file on communication channel

[shrikant_1]

Hi all,
I am working on HP-UX and handling 5 oracle 9i databaes. I have migrated one oracle8 database to oracle9i database. But users of that database are complaining that their sessions are getting disconnected by ora-03113. Also, some commands like update are taking lots of time or sql-plus is hang by this commands sometimes.
Any reply will be greatly appriciated.
Thanks
shrikant

[Peter Godron]

shrikant,
as you probably know by now 03113 is a 'generic' error i.e. a catch-all.
Do a quick check of the OS, any i/o problems.
Check the alert.log as well.

Ensure that you have enough system resources to fulfill the oracle requirements (SGA, etc). Then start one database and check proper startup. Then check the listener and any http server connections.
Finally allow a limited number of users and check resources whilst running.
We have found problems in the past from corrupted dbs (i/o errors) to flooded listeners.

Regards

[Jean-Luc Oudart]

ORA-3113=> The connection to the server was lost.

Usually other error messages associated

see Metalink doc id 17613.1
It explains where to look , what to do
(lots of scenarii)

Regards
Jean-Luc

[Indira Aramandla]

Hi Shrikant,

As mentionned in the earlier posts â ORA-3113â is a very generic error.
An ORA-3113 "end of file on communication channel" error is usually reported by a client process connected to an Oracle database. The error basically means 'I cannot communicate with the Oracle shadow process'. As it is such a general error more information must be collected to help determine what has happened - this error by itself does not indicate the cause of the problem. For example, ORA-3113 could be signalled for any of these scenarios:
This error can be caused by just about any server-side error and also network errors. The error simply means you were disconnected for some reason. The best thing to do is first go through your database alert log (background_dump_dest) and see if any errors/messages are being logged. Also check for more specific errors in trace files generated in user_dump_dest.
You have 5 (9i) instances on the server. And you migrated one instance form 8i to 9i. Was the migration successful. Are you connecting to an older version of Oracle database? When did you first start receiving the error?

It is common for this error to be accompanied by other errors such as:
· ORA-1041 internal error. hostdef extension doesn't exist
· ORA-3114 not connected to ORACLE
· ORA-1012 not logged on
· ORA-07445: exception encountered: core dump
These are all symptomatic of being disconnected from Oracle.

As Jean mentioned Note:17613.1 in Metalink â ORA-03113 on Unix - What Information to Collectâ provides more analysis

[Yogeeraj_1]

hi shrikant,

" update are taking lots of time or sql-plus is hang "

I had a recent case of a similar problem whereby the root cause was an application with was not using Bind Variables properly and consuming loads of resources..

We usesd Staspack and SQLTrace to trace that application and the database environment.

hope this helps!

regards
yogeeraj