Hi
We are maintaining a huge LDAP Directory with x thousands records. This OpenLDAP instance is only for load and performance tests of our software. I've recently noticed that the bin logs of the bdb backend are growing nearly unlimited. (The LDAP Server gets replicated to an other one in an active/passive cluster). To gain the "wasted" space back, we're writing the bin logs back to to the main database every 24 hours. Now we are faced with an ldap server which can only read from its database. After an restart of the server everything works as expected but the same happens again irregularly...
My Questions are:
1. what could happen with the server, which makes it only being able to read of its own db? 2. why are the bin logs are growing so dramatically?
Thanks for any help Tom
On Thu, 12 Jul 2012, Thomas Spycher wrote:
[...]
I've recently noticed that the bin logs of the bdb backend are growing nearly unlimited.
[...]
- what could happen with the server, which makes it only being able to
read of its own db? 2. why are the bin logs are growing so dramatically?
I'm not entirely sure what you're referring to when something is only "able to read of its own db" (is the backend going read-only? what messages are you seeing? etc.) but given the context perhaps you're experiencing timeouts from a large log sync. Berkeley DB transaction logs grow without bound under the default configuration, and you might not be checkpointing writes in back-bdb (which could explain your curious restart behavior). Get started with:
http://www.openldap.org/doc/admin24/maintenance.html http://www.openldap.org/faq/data/cache/738.html
and the slapd-bdb man page for details on transaction logs, checkpoints and so forth. You might also want to check that you're running a sane version of Berkeley DB (with patches as may be appropriate) and of OpenLDAP itself (2.4.31 is latest as of this writing).
openldap-technical@openldap.org
-
Aaron Richton -
Thomas Spycher