• Article Type: General
• Product: Aleph
• Product Version: 16.02

Description:
As of this morning, we began seeing Oracle errors

ORA-12571: TNS:packet writer failure,

and

Oracle error: update_cursor z11
ORA-03114: not connected to ORACLE

in the oclc log file. Records we exported from OCLC were not showing up in Aleph. We restarted the server and that fixed the problem - at least for the time being. I want to be sure this doesn't happen again. What do the errors mean? How can we prevent this?

[Note: The same messages could appear in the ue_01 log.]

Resolution:
This could indicate that Oracle isn't running. (Do util o/3 to check.)

If the Oracle server is running, it may be that Oracle came down while the process was still running. (ALEPH Shutdown did not occur or did not complete.) Oracle could be brought down for a backup or it could come down unexpectedly. (The log indicates that this process has been active since yesterday.) The aleph_shutdown should be run before doing the backup and this script should be shutting down the oclc_server -- and the other servers.

Perhaps you should follow this principle: If Oracle comes down unexpectedly, restart the servers (and daemons). If it was unexpected, then it shouldn't happen very often.

Also found this on Web about ORA-12571:

Possible causes and remedies:
With Oracle 8.0.4 onwards, this error may indicate that the database is in the process of being shut down.
This error has been observed when there was a loose cable connection to the SCSI disk array (on an Oracle Parallel Server system). Related errors were also subsequently noted in the NT event log. If this is the cause then the remedy is to check all the cable connections. Do this by hand rather then visually, since a cable not being firmly in place could cause the problem.
Daniel Haller Strahm has reported that this problem can also be caused by an IP address conflict (Oracle 9.0.1). Windows will normally log an IP conflict in the event log and show a pop up alert - but only if it detects the conflict. Check that no other computer on the network has the same IP address as the machine on which the problem has been reported.
<end Web citation>