Re: Segfault in slapd 2.4.8 - syncrepl and glue (ITS#5430)

[hyc@symas.com]

Duncan Gibb wrote:
> hyc@symas.com wrote:
> HYC>  syncrepl.c is now updated in HEAD to allow this replication to occur.
>
> HYC>  Using your test, I don't see any crash.
>
> Thank you.  We'll re-test with current HEAD and post the outcome.

I modified your test case to make it easier to run. The new test is in 
ftp.openldap.org:incoming/its5430.tgz.1

You simply extract it inside the tests/ subdirectory of the build tree. After 
you have built the OpenLDAP source and run "make test" you can cd into the 
its5430 directory and run the test.sh from there.

There were also some errors in your config file templates, which are corrected 
in the above archive.

Most notably, in template 1, you explicitly configured syncprov on top of glue 
instead of taking the default behavior of syncprov under glue. As documented 
in slapd.conf(5), you should only do that when you want a single syncprov 
overlay to be master of the entire tree. In this case, you clearly want 
syncprov to only master one portion of the tree, with the subordinate being 
taken care of by its own syncprov overlay. With your incorrect configuration, 
server3 would receive redundant updates because the subordinate branch would 
be propagated twice.

You also inserted the overlay directives in the middle of your database 
clauses, which is poor practice (and caused crashes in older OpenLDAP 
releases). Again, as documented, overlays are meant to be configured at the 
end of their respective databases.

Not an error, but you used unique rids across all of the config files. rids 
only need to be unique within a single server.
-- 
   -- Howard Chu
   Chief Architect, Symas Corp.  http://www.symas.com
   Director, Highland Sun        http://highlandsun.com/hyc/
   Chief Architect, OpenLDAP     http://www.openldap.org/project/