[Replicator-users] Working repli-sync [was: Tests for repli-sync ?]

[Bernd H. <bj...@da...>]

Hello,

first of all, I should apologize. In the repli-sync test I report on
Fri, 25 Oct 2002, I forget the --real param. After working around some
bugs I have simply forgotten it.

To prevent this in futur and to help others using repli-sync I write a
small howto. You find it attached in the formats PDF and LyX.

At the End of this document you find a list of bugs I found and a list
of proposals for the futur of replicator. The howto must be corrected
after fixing the mentioned bugs, but then I like it t be included in
replicator.

What do you think about that?


On Thu, Oct 31, 2002 at 10:15:16PM +0100, Sebastien Chaumat wrote:
> The "-text2" is superseeded by the future -t$fs :)
> (I though I did commit this... maybe it still on my hd)
> Can you check, in documentations, if "mkfs -t fs" works whatever fs is?
> I mean is there any filesysteme that do not follow the "mkfs -t"
> convention. 

Looking in the manpage I see that the following types are supportet:

       -t vfstype
              The argument following the -t is used to indicate  the  file  system  type.
              The  file  system types which are currently supported are: adfs, affs, aut­
              ofs, coda, cramfs, devpts,  efs,  ext2,  ext3,  hfs,  hpfs,
              iso9660,  jfs, minix, msdos, ncpfs, nfs, ntfs, proc, qnx4, reiserfs, romfs,
              smbfs, sysv, tmpfs, udf, ufs, umsdos, vfat, xfs.

I had some problems with the "autofs" type in the past. Sometimes (on
floppies) the filesystem was not correctly detected. I think you should
not use this option.


> The point is that repli-update need cleaning. I want to use standard
> argument parser (as in repli-miniroot).

Is this important enough to delay the first release for woody or can it
wait for the relase after that.


>  The folowing maybe interresting :
> 
> split replicator.deb in 3 :
> 
> replicator-miniroot-server : only repli-miniroot 
> replicator-model-server : only repli-bootdisk and something to get info
> about the model (I guess maximum info about the model should be gathered
> at replication time only via rsync)
> replicator-target-syncer : repli-sync

I do not understand what the goal of this is. Replicator is a small
tool and it does no harm, I think, to have all tools present on all
hosts. Nevertheless this makes it nessesary to prevent
"replicator-*-server" from beeing synchronized to the targets.
I feel, this is less important extra work, but maybe you can explain it
a bit.


What do you think are the things, that should be done for the first
woody release and what do you plan after this?

My proposal is:

1.) Make a list of the known bugs. (e.g. the ones listed in my
    attachment and in the buglist on sourceforge.
2.) Fix this bugs.
3.) Update documentation.
    I can provide an (update) howto for the first replication
    (repli-install) and repli-sync.
4.) Release replicator for woody.
5.) Collecting wishes for the next release.

What do you think about this? I like to help on all points.


Regards,
Bernd