View this page "Suggestions for re-syncing"

5 views
Skip to first unread message

jason

unread,
Nov 27, 2007, 8:53:36 PM11/27/07
to chironfs-forum


Click on http://groups.google.com/group/chironfs-forum/web/suggestions-for-re-syncing
- or copy & paste it into your browser's address bar if that doesn't
work.

Vuco

unread,
Nov 29, 2007, 12:24:18 PM11/29/07
to chironfs-forum
This idea is really cool and very close to the ones I was discussing
with Alex ,
member of this forum, and Infwonder, from Zenwalk Linux.

The current state of the brainstorm is to let the user mount ChironFS
passing
the path to the resync-script in the command line. The resync script
will be
called by ChironFS with an open pipe.

So, ChironFS will interact with the script through its stdin/stdout
offering
commands to it like many tcp/ip services do on a network connection
(http, smtp, pop and imap for example).

The script will be able to:

1) list all failed writes

2) provide the resync itself and tell ChironFS what exactly was
synced

3) tell ChironFS to resync something (all replicas, one replica, one
operation in one replica)

4) change the status of a replica. Suggested examples are: disabled
(it failed and not accessible, ChironFS will avoid any access to it),
enabled (replica is accessible and integrity of the data is 100%),
syncing
(the replica is accessible, but data is not 100% equal to the enabled
replicas,
ChironFS will access only the data it knows that is goodin read only
mode),
caching (the replica is accessible, but data is not 100% equal to the
enabled
replicas, ChironFS will access only the data it knows that is good and
will write
any new data for it).

5) tell ChironFS to block all write operations

6) tell ChironFS to block all operations

7) tell ChironFS to disable blocking

Doing so, the user will be able to set its own resync rules. Of course
some ready
to use scripts will be available. I thought in at least 2 scripts:

1) a simple one that operates blocking ChironFS, doing all resync in
mount time
and enabling the replica at the end

2) a pair of master-slave resync scripts, comunicating over network,
enabling a
centralized coordinated resync

So, this is what is being planned for now. Any comments/ideas?

Thank you!

Luis Otavio de Colla Furquim

Vuco

unread,
Nov 29, 2007, 1:09:54 PM11/29/07
to chironfs-forum
Sorry about the subject changed including portuguese words.
It was a mistake.

I am trying to fix it.
Reply all
Reply to author
Forward
0 new messages