OpenLDAP service on RHEL5.5 must be restarted after large burst of connections

Issue

• We had an instance where several of our OpenLDAP slave servers running RHEL5.5 stopped responding to LDAP requests.
• Several of the slaves thought there were too many replication requests in-flight and began rejecting connections.
• 'ldapsearch' against the problematic hosts was failing, though as best I can tell from the logs slapd was still running.

• A 'service ldap restart' of the problematic servers fixed our issue.

  2010-07-03T08:57:47.959420-05:00 HOSTNAME slapd[11016]: connection_read(61): no connection!
  2010-07-03T08:58:12.668799-05:00 HOSTNAME slapd[11016]: connection_input: conn=63677 deferring operation: too many executing
  2010-07-03T08:58:12.668832-05:00 HOSTNAME slapd[11016]: connection_input: conn=63677 deferring operation: pending operations