corosync/TODO

The corosync standards based cluster framework TODO list
Last Updated: May 26, 2006

P1 items are to be implemented before Wilson release.  P2 items may be
implemented in Wilson if interested parties provide patches.  P3 items are
targeted for Humphrey.

Generic Items
-------------
* P1 32/64 bit cross-endian must be working.
* P2 Itemize any changes required for AIS B.02.01 in the TODO list.
* P2 doxygen-ize the include and lib directories.
* P2 There are many TODOs in the code that need attention.
* P2 Implement static voting virtual synchrony filter.
* P3 Integrate with rgmanager system.
* P3 doxygen-ize the exec directory.
* P3 Advanced Synchronization Engine needed to synchronize data without
     long blocking delays during configuration changes.

Totem
-----
* P1 Test scalability to 128 processors.
* P1 Implement safe message delivery (for both Wilson and Picacho).
* P1 Add mechanism to reenable a redundant ring after that ring has been
     declared faulty and then repaired by the system administration.
* P1 In redundant ring configuration disallow binding to localhost interface.
* P2 Flush totem messages during RECOVERY state that are in the
     new_message_queue.
* P2 Turn totem layer into an LCR component.
* P3 Implement the totem multiring protocol.

YKD Virtual Synchrony Filter
----------------------------
* P1 32/64 bit cross-endian must be working.
* P2 Scale to 128 processors.

LCR
---
* P2 Finish live component replacement services.

Checkpointing
-------------
* P1 32/64 bit cross-endian must be working.
* P1 There is a bug in that iteration doesn't work under heavy load that
     must be fixed.
* P1 Creation/deletion of checkpoints should be done using SAFE messaging.
* P1 The checkpoint unlink operation doesn't work as per specification.  The
     specification says that a checkpoint unlink should then allow a new
     checkpoint with the same name to be created.  All cluster members using
     the old checkpoint continue to use the old checkpoint until they all
     stop using the checkpoint, at which point the checkpoint should be removed
     from the cluster. 
* P2 Conformance testing via SAF-TEST needs to reach 100%.

Distributed Locking Service
---------------------------
* P1 32/64 bit cross-endian must be working.
* P1 Distributed locking needs configuration change support.
* P1 Provide kernel-based DLM service handler for distributed locking.
* P2 Conformance testing via SAF-TEST needs to reach 100%.

Messaging Service
-----------------
* P2 32/64 bit cross-endian must be working.
* P2 Finish implementation.
* P2 Conformance testing via SAF-TEST needs to reach 100%.

Availability Management Framework
---------------------------------
* P1 32/64 bit cross-endian must be working.
* P1 Finish next generation state machine design.
* P1 Implement next generation state machine design.
* P2 Conformance testing via SAF-TEST needs to reach 100%.

Logging Service
---------------
* P2 32/64 bit cross-endian must be working.
* P2 Design and implement.
* P3 Conformance testing via SAF-TEST needs to reach 100%.

IMMS
----
* P3 32/64 bit cross-endian must be working.
* P3 Design and implement morphing the configuration object database.
* P3 Conformance testing via SAF-TEST needs to reach 100%.

Cluster Membership
------------------
* P1 32/64 bit cross-endian testing.  Code completed and believed to be correct.

Eventing
--------
* P1 32/64 bit cross-endian must be working.

Closed Process Groups
---------------------
* P1 32/64 bit cross-endian must be working.

Extended Virtual Synchrony Passthrough Interface
------------------------------------------------
* P1 32/64 bit cross-endian must be working.
* P1 Modify man pages to match new API semantics.
* P1 Add mechanism to get redundant ring information from EVS interface.

IPC
---
* There are no TODOs for the IPC system.

SYNC
----
* P1 32/64 bit cross-endian must be working.