I had been putting together a small list of items I think would be nice to
have in user documentation (a wishlist, if you like):
1) Installing Sheepdog and it's dependencies:
* Choosing the appropriate cluster manager (discuss corosync, zookeeper
and accord and the pros and cons of each)
* Packages or source code? (Install from .debs, Install from .rpm,
Compiling from source)
2) Modes of operation
* Different ways of running sheepdog ('conventional' cluster, storage-only
cluster, qemu-host-only cluster, etc)
* Single node cluster (developer/testing mode)
* A cluster with one sheep per physical host
* A cluster with several sheep per node (physical disk dedicated to each,
using zone IDs to preserve redundancy)
3) Running your cluster
* Starting and stopping your cluster safely
* Running, testing and monitoring your cluster
* Common issues and how to deal with them
* Sheep command line options explained (performance, safety, etc)
4) Examples and advanced stuff
* Sheepdog with OpenStack
* Sheepdog with CloudStack
* Sheepdog with Ganeti
* Building a large cluster - cluster network considerations, configuration
management, running diskless nodes with all storage dedicated to sheep.
Some of this is wishful thinking but how does it sound? - put yourself in
a newbie's shoes :-) I can start this off and welcome input from others
(please let us know what you are doing so we don't duplicate work).
Kazutaka, can we have a sheepdog-users list please? -either at
lists.wpkg.org, or somewhere else suitable.
Thanks,
Matt.
|
|