On 1/14/11 10:37 AM, Pierre-Arnaud Marcelot wrote:
On 14 janv. 2011, at 10:23, Emmanuel Lecharny wrote:
On 1/14/11 9:58 AM, Pierre-Arnaud Marcelot wrote:
Hi Emmanuel,
I think it should.
In the current state, does this mean that any modification done on ou=schema
will not be saved and will be lost if the server is rebooted?
No, modifications done on ou=schema are persisted on disk. Only the
modifications done on cn=schema are not persisted, AFAICT.
Hum... I'm confused...
You just said the opposite in your first mail: "[...] modification is not stored on
disk in ou=schema"...
yes, this is what I meant. When you do a modification in cn=schema (in
memory), it's not stored on disk in the ou=schema partition.
I'll create a JIRA and a test to demonstrate the issue.
Fixing it should not be a problem, it's just a a matter of converting the
schema element (which is passed using the schema element syntax) to a
meta-schema entry, and propagate it to the backend.
Remember that modifications to cn=schema are *not* allowed (it's a read only
data structure) from outside the server, but it's always possible to modify the
rootDSE subschemaSubentry attribute, as it contains all the loaded schema
element. This will, in fact, impact the cn=schema, as it's just a LDAP
exposition of the loaded schema.
Hum... I'm confused again...
To my knowledge, 'subschemaSubentry' attribute value points to the "cn=schema"
DN and that's in this particular entry that you can access the schema elements (via
'attributeTypes', 'comparators', [etc.] attributes).
I'm 100% sure 'subschemaSubentry' attribute does not contain any loaded schema
element.
The rootDSE entry contains the subschemaSubentry AT, which contains a
reference to the cn=schema virtual partition.
You can modify the cn=schema elements by adding for instance things like :
( 1.3.6.1.4.1.18060.0.4.1.2.10000 DESC 'bogus desc' SYNTAX 1.2.3.4
X-SCHEMA 'nis' )
Sorry for the confusion.
--
Regards,
Cordialement,
Emmanuel Lécharny
www.iktek.com
