Oidldapd Log Fills Up With Error: DispatcherListener: Unable to accept New TCP connection (Doc ID 275712.1)

Last updated on MARCH 08, 2017

Applies to:

Oracle Net Services - Version 8.1.7.4.0 and later
Oracle Internet Directory - Version 9.0.4 and later
Information in this document applies to any platform.

Symptoms

Oracle Internet Directory (OID).

OID dispatcher log, i.e., oidldapd01.log, fills up with "DispatcherListener: Unable to accept New TCP connection" errors, consumes all the available disk storage, then crashes OID (Oracle Internet Directory) and the server.

Excerpt of sample log:

2004/03/19:02:00:39Starting up the OiD Server, on node ldap2
2004/03/19:02:00:39Oid Server Connected to DB store via iasdb.<[domain]> connect string.
2004/03/19:02:00:39OiD LDAP server started.
2004/03/19:02:00:42DispatcherListener: Unable to accept New TCP connection
2004/03/19:02:00:47DispatcherListener: Unable to accept New TCP connection
2004/03/19:02:14:54DispatcherListener: Unable to accept New TCP connection
2004/03/19:04:05:49DispatcherListener: Unable to accept New TCP connection
2004/03/19:04:19:56DispatcherListener: Unable to accept New TCP connection
2004/03/19:04:21:57DispatcherListener: Unable to accept New TCP connection


These files, if not deleted regularly, can exceed into the GBs.

Scenario:
Cisco load balancer (LBR) host (named ldap for example)
oid host1 (named ldap1) - full 9iAS or 10g infrastructure
oid host2 (named ldap2) - OID only (typical install via Oracle Installer; not custom install)

All of a sudden, about 1-20 of the error message can be generated every ten seconds, and within a week the log files can grow to the point where the server crashes.

Nothing else running on these boxes; no midtiers associated with this infrastructure.

Not using sso for portal because everything needs to reside in a single database (would have to use h/w failover which is too expensive).

Using Cisco load balancer for reading only, writing always occur against one server (ldap1) and then gets replicated to ldap2.

There is hardly any traffic from a query standpoint, and very few users.

Running a tool called Siemens Directs Metahub, metadirectory synchronization tool, but problem occurs sporadically, and it doesn't seem to have any correlation to the problem. It reads from an HR system and populates OID with data.

There is also a workflow running against only ldap1 server, which does a lot of reading - the entire directory every night; about 5000 entries in the directory - but changes occur during the day.

These errors messages appear on both servers, but on ldap1 is severe enough to cause server/OID crashes.

Similar issues were reported to Oracle Support in the past, and the following was reported as solution:
"Removing the Cisco load balancer and a second ethernet card out of the picture, assuming that it may be flooding us with keepalive packets, and plugging the box with only one ethernet card directly into the router, the error goes away.

A google.com search for "Unable to accept New TCP connection" returns many hits, which lead us to look at our router and load balancer which was sending and not acknowledging KeepAlive messages to the port that oidldapd is listening on."

Changes

 

Cause

Sign In with your My Oracle Support account

Don't have a My Oracle Support account? Click to get started

My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms