|
Size: 1475
Comment:
|
Size: 2194
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 7: | Line 7: |
* [[Roadmap]] === Development Tools === * [[Cluster Design GUI/Integrated Development Environment]] * [[Source Code Structure and Makefile System]] * [[Install and Build]] |
|
| Line 16: | Line 24: |
| * [[Dictionary]] | * [[Checkpoint]] (AKA Dictionary, map, or hash table) |
| Line 18: | Line 26: |
| * [[ResourceAssociatedKeyDictionary]] This [[Dictionary]] associates a ClusterUniqueId to each key and calls a cleanup whenever the resource represented by the ClusterUniqueId disappears (due to a fault). |
* [[ResourceAssociatedCheckpoint]] This [[Dictionary]] associates a Handle to each key and calls a cleanup whenever the resource represented by the Handle disappears (due to a fault). * [[ThreadPool]] A group of threads that are applied to a list of tasks. * [[MessageFlowManager]] This class intermediates between the messaging layer (send/receive) and the application code. It manages messaging buffers and optimizes frequency of message sending verses message size. |
| Line 24: | Line 37: |
| Interface with SAFplus 6.x logging implementation underneath | Applications and other services can log to shared memory. A log server will "scrape" the shared memory and post the message to log channels and/or files. |
| Line 26: | Line 39: |
| Interface with SAFplus 6.x IOC messaging implementation underneath, but takes a ClusterUniqueId as a destination | Interface with SAFplus 6.x IOC messaging implementation underneath, but takes a Handle as a destination |
| Line 28: | Line 41: |
| Group membership and master election | Group membership and cluster master election |
| Line 30: | Line 43: |
| String name to arbitrary data (generally a ClusterUniqueId) | String name to arbitrary data (generally a Handle) |
| Line 32: | Line 45: |
| All entities report errors to this centralized fault manager. Only it chooses to fault an resource identified by ClusterUniqueId |
All entities report errors to this centralized fault manager. Only it chooses to fault a resource (node or process). * [[Availability Management]] Core process that starts and monitors applications * [[Remote Procedure Call]] Fast and efficient RPC subsystem |
SAFplus 7 Design Specification
- This chapter describes general design requirements and rules-of-thumb to ensure a consistent semantic and API style across all services.
Programming Style Requirements
- This chapter describes programming style so that SAFplus provides a consistent look.
Development Tools
Fundamental Objects
- This is a cluster-wide identifier
Derives from HandleT, adds a cluster identifier to become globally unique
- Cross object, node, cluster, transaction
Checkpoint (AKA Dictionary, map, or hash table)
- Replicated, nested, persistent, notifiable, shared memory, transactional Hash Table
This Dictionary associates a Handle to each key and calls a cleanup whenever the resource represented by the Handle disappears (due to a fault).
- A group of threads that are applied to a list of tasks.
- This class intermediates between the messaging layer (send/receive) and the application code. It manages messaging buffers and optimizes frequency of message sending verses message size.
Services
Logging Applications and other services can log to shared memory. A log server will "scrape" the shared memory and post the message to log channels and/or files.
Messaging Interface with SAFplus 6.x IOC messaging implementation underneath, but takes a Handle as a destination
Group Group membership and cluster master election
Name String name to arbitrary data (generally a Handle)
Fault All entities report errors to this centralized fault manager. Only it chooses to fault a resource (node or process).
Availability Management Core process that starts and monitors applications
Remote Procedure Call Fast and efficient RPC subsystem