Hi Micha On 01.04.2012 11:03, Michael Bell wrote:
I am a little bit confused. First you want to update the configuration in the parent process and second you think that it is not a good idea to place all the data in the parent process (the owner of the "root" context). Is the connector style config something like a configuration server?
The connector should (in the first step) provide two things:
1) read only access to the "inner configuration" of the system. Where "inner configuration" can been translated to something like a basic configuration which is once setup and determines the basic operation of the system. This config parameters are most likely in the local config files (which is the XML stuff in the old config system)
2) read only access to the "outer configuration". These are configuration items that refer to "things" brought into the system from outside, e.g. specify the key size, default puk, etc of different smartcard families.
3) Read and Write access to user data, as reading and publishing certificates using LDAP.
The Connector behaves towards the system as a single tree. Towards the "data" you can define arbitrary nodes to behave as a "Proxy" and use the remaining branch as parameters to do a LDAP, database, whatever query.
What I tried was to create a kind of "overlay" to 1 to build crossref-structures. Technically I used a read/write Proxy connector that just stores all info in a perl hash into its memory. The object reference to that Proxy Connector is stored inside the basic connector (as we now found out only in the forked child).
The basic question would be: Can we manage a cache that spans over all childs and is updateable from inside the childs or would it be easier to create such cross-ref structures inside root and copy it to each child. I am am totally unaware of what perl does when forking and I dont now how expensive such support would be.If we update the "root" context then every fork creates a copy of this information. I see no difference according to pre-initialization on startup. We could create a queue which only stores a specified maximum number of cross references but what is a good limit? This depends on the capabilities of the server and the use case.
Oli -- Protect your environment - close windows and adopt a penguin! PGP-Key: 3B2C 8095 A7DF 8BB5 2CFF 8168 CAB7 B0DD 3985 1721
smime.p7s
Description: S/MIME Cryptographic Signature
------------------------------------------------------------------------------ This SF email is sponsosred by: Try Windows Azure free for 90 days Click Here http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________ OpenXPKI-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/openxpki-devel
