[sheepdog] [PATCH 0/4] restore default redundancy for cluster snapshot

Hitoshi Mitake mitake.hitoshi at gmail.com
Mon Mar 24 16:03:34 CET 2014


At Mon, 24 Mar 2014 23:47:29 +0900,
Hitoshi Mitake wrote:
> 
> At Mon, 24 Mar 2014 23:43:21 +0900,
> Hitoshi Mitake wrote:
> > 
> > At Fri, 14 Mar 2014 20:04:36 +0800,
> > Liu Yuan wrote:
> > > 
> > > As previously discussed, after
> > >  $ dog cluster snapshot load ./path
> > > 
> > > we will lose the default redudancy in the previous cluster and simply get '3'
> > > set for '-c'.
> > > 
> > > This patch kill this problem by storing cluster state in the header of snap log
> > > file.
> > > 
> > > Liu Yuan (4):
> > >   dog/farm: kill snap file
> > >   sheep: add a operation to get cluster info
> > >   dog/farm: add snap log header
> > >   dog: restore default redundancy scheme for 'cluster snapshot load'
> > > 
> > >  dog/cluster.c            | 10 ++++-
> > >  dog/farm/farm.c          | 59 +++++++++++++++++++++---------
> > >  dog/farm/farm.h          | 21 +++++++----
> > >  dog/farm/snap.c          | 95 ++++++++++++++++++++++++++++++++++++++----------
> > >  include/internal_proto.h |  1 +
> > >  sheep/ops.c              | 15 ++++++++
> > >  6 files changed, 154 insertions(+), 47 deletions(-)
> > 
> > Looks good to me.
> > Reviewed-by: Hitoshi Mitake <mitake.hitoshi at lab.ntt.co.jp>
> > 
> 
> BTW, please add a new test or enhance existing 030 for testing this
> feature later.

One more request: please add a description about the format change of
cluster snapshot to CHANGELOG.md.

Thanks,
Hitoshi



More information about the sheepdog mailing list