neo4j backups in AWS

[Scott Muc]

Hi neo4j peeps,

I was considering doing EBS snapshots as a method to backup our neo4j instance that's in EC2. Will this work while the server is live or does some server downtime need to be orchestrated to do this.

Scott

[Peter Neubauer]

Scott,
The safest way to do it without HA is to shutdown the db, thus flushing any transactions to disk, so I think you should do that.

[Nikhil]

Scott,

I'll second on Peter's suggestion. We have added a cron job to restart the server once in a day (traffic trends drive this schedule). Taking an EBS snapshot helps taking a backup when you're not on HA.

Another way to do it would be via GraphML/GraphSON. But that needs some sanity checks when you're adding nodes/edges to your graph. You might run into de-serialization issues while restoring from a GraphML/GraphSON.

Hope this helps.

--

Nikhil

--
Nikhil Lanjewar
Engineering Lead at YourNextLeap
http://yournextleap.com

[Scott Muc]

Thanks guys

Since we're just ramping up and don't need 5 nines of uptime I think
the cronjob that stops the server and performs an EBS snapshot is
probably the simplest thing for us to do.

Scott

On May 2, 1:05 am, Nikhil <nikhil010...@gmail.com> wrote:
> Scott,
>
> I'll second on Peter's suggestion. We have added a cron job to restart the
> server once in a day (traffic trends drive this schedule). Taking an EBS
> snapshot helps taking a backup when you're not on HA.
>
> Another way to do it would be via GraphML/GraphSON. But that needs some
> sanity checks when you're adding nodes/edges to your graph. You might run
> into de-serialization issues while restoring from a GraphML/GraphSON.
>
> Hope this helps.
>
> --
> Nikhil
>

> On Wed, May 2, 2012 at 11:03 AM, Peter Neubauer <neubauer.pe...@gmail.com>wrote:
>
> > Scott,
> > The safest way to do it without HA is to shutdown the db, thus flushing
> > any transactions to disk, so I think you should do that.

[Peter Neubauer]

Cool.
Let us know how it goes!