Re: SEGV on syncRepl provider (ITS#3296)

[richton@nbcs.rutgers.edu]

Owing to undesirable retry characteristics in earlier releases, we have
since changed to refreshOnly. Full slapd.conf can be provided but
configuration overview is one syncRepl provider with two syncRepl
refreshOnly slaves, all running OpenLDAP 2.2.18 with associated software
at same revisions as originally described. The provider stores six HDB
databases ("dc=X,dc=rutgers,dc=edu") and the consumers store six
identically named databases, with the "subordinate" keyword applied under
"dc=rutgers,dc=edu".

The syncrepl definition, for reference, where "X" varies with leading dc:

syncrepl rid=10X
        provider=ldap://stanlee.rutgers.edu/
        starttls=critical
        type=refreshOnly
        interval=00:00:05:00
        searchbase="dc=X,dc=rutgers,dc=edu"
        scope=sub
        updatedn="cn=syncReplConsumer,dc=X,dc=rutgers,dc=edu"
        bindmethod=simple
        schemachecking=on
        binddn="cn=readEverything,dc=X,dc=rutgers,dc=edu"
        credentials=credentials


I will work on producing provider logging (d=255 to start?) of a crash.
local4 syslog information is available at all times from all systems; for
instance, I can look back to find the last few operations to the
backtraces that have been attached to this ITS.