CORAL Server Meeting

Friday 23 May 2008, 14:00 → 15:00 Europe/Zurich

31/2-029 (CERN)

Design and development issues of the CORAL server project, and integration with the ATLAS dbproxy project.

14:00 → 14:20 Coral Server Protocol modifications

See the atteched PDF in Indico, modifications are highlighted (or in the usual place)

paper CoralServerProtocol.pdf

Clarified the meaning of the cacheable flag: - The cacheable response messages do not need the cache key, we need to generate only for the requests - Only the cacheable responses for the cacheable requests are cached. - All the response messages should be cacheable except the Exception message: - by default, it is cacheable. - the server may decide if an exception message is cacheable or not. The exception message have variable cacheability property.

14:20 → 14:30 Defining the client monitoring information subprotocol

Alexander defines the sub-protocol. It will include 4 fields:

Process ID ¦ Last 32 bytes of command name ¦ Timestamp ¦ Username

We need to define a separator and the order of fields, and add to the protocol description document.

This information is in the Connect message, an Array field after the key. It is OK, if the proxy need to analyze this field as well.

14:30 → 14:40 Server/client/proxy implementation status

- The key generation for the cacheable messages are implemented - The query codec development is ongoing - Adaptation of the server code to the query codec protocol is ongoing

14:40 → 14:50 Plans and actions for the next 2 weeks

- Zsolt is off for 2 weeks, but reads mails regularly and may give support if needed - Alexander will: - finish the query codec implementation in the server code - propose a simple protocol for the client monitoring info, and implement it - extend the query codec toward more complicated queries

14:50 → 15:00 AOB

- There will be a new CORAL patch release for Atlas (v 1.9.7)