ODSEE - Directory Server Stops Responding to Connections and in Some Cases Can Recover After a Few Minutes (Doc ID 1388346.1)

Symptoms

Symptoms can include one or more of the following:

Changes

The cause is due to an excess buildup of Bad/Stale SSL connections that can lock the Directory Servers threads.    In some cases, it can clear up on its own.

To determine if you are hitting this issue you can use the following methods:

        1.) If a DirTracer is taken WHILE the issue is happening you can use the cn=monitor searches from the DirTracer capture.   For example, in one case we could see there are 176 LDAPS (SSL) connections in the "cn=monitor" output:

[/<DirTracer/capture>/ds-srch-monitor]#grep -c LDAPS *
monitor.111215-115714:176
monitor.111215-115721:176
monitor.111215-115726:176
monitor.111215-115732:176
monitor.111215-115738:176

        2.) In the example below, DirTracer also took a series of pstacks where there are SSL handshakes.  Reference the SSL functions in the pstack summaries example below.

Using a series of pstacks, you can visually scan the threads for all functions that begin with the "ssl" or "ssl3."  You can also use the attached Pstacker script to create pstack summaries of the captured pstacks. 

Pstacker, extracts the functions from pstack threads and group the like threads functions together.  Each line contains the number of like threads the pstack had.  The final line shows the total number of threads the process has.

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.