lic_gma at hotmail.com
Thu Jun 28 11:48:25 EDT 2012
If I might add up to it, I would like to know specifically what would be
a proper way to have a replication site of a cluster, it seems to me it
is easier to backup the replicated site than the cluster itself.
We replicate using a Java scheduled job but it requires us to mark each
key with an specific index (lets say; the version of the key), so we
know what version we were left at on the previous copy and then we
continue where we were before to the current-1 version.
If we could replicate the cluster in a sort of a site-to-site
replication, increasingly, then I would think it will be the perfect
solution as it will be way easier to backup from the replicated site.
On 28/06/12 16:37, Shane McEwan wrote:
> I was about to ask this exact same question so I'll add my specific
> queries to the conversation, if I may.
> In particular, we're using LevelDB and are considering filesystem
> snapshots and then syncing the snapshots off-site. But what happens if
> the snapshot is taken while LevelDB is in mid-write? Would LevelDB
> cope if that snapshot was restored as the live database?
> On 28/06/12 14:15, cvalue at yandex.by wrote:
>> I've got one little question: what is the fastest and safest way to
>> backup Riak cluster? Should I use riak-admin tool or pack base and
>> it's config manually into tarball?
>> Thanks in advance,
>> riak-users mailing list
>> riak-users at lists.basho.com
More information about the riak-users