[okfn-help] Backup work

Mon Jun 7 20:16:32 BST 2010

On 7 June 2010 18:08, James Casbon <casbon at gmail.com> wrote:
> Apologies for the delay in getting back to you on this, Rufus.

No problem :)

> On 25 May 2010 15:21, Rufus Pollock <rufus.pollock at okfn.org> wrote:
>> Dear James and Martin,
>>
>> I'm still not quite sure about the state of your guys efforts on the
>> backup front. As I missed you guys last night I had a stab on my own
>> at taking things forward.
>>
>> 1. I updated this document to reflect the current situation (as I
>> deciphered it):
>>
>> <http://knowledgeforge.net/okfn/tasks/wiki/BackupPlan>
>>
>> 2. I updated fabfile.py backup_report and started on backup_setup method.
>>
>> At this point 'd be really grateful :) for some info on what machines
>> you've (starting) setup backup on already (and current status). I
>> would also like thoughts on:
>
> I don't have this info (martin was doing it), which is why I started
> writing the report on the backup status of the machine so that I could
> get this info automatically.  I left this comment on the ticket:
> """
> Nothing on eu3, eu4, eu1.

Ok, so those are the priority.

[...]

>>  * s3 snapshots of ebs backups
>
> I think that the main problem we might face is a long downtime in a
> particular availability zone that makes both backup and host
> unavailable.  However, at this point amazon would be loosing a lot of
> cash so I think they should bring it up quickly enough.

What about the fact that EBS might fail? The reliability of EBS is
reasonable but not amazing [1] (of course we'd need both main storage
and the EBS to fail in the same time period to have issues). S3 is
much reliable and is supposed even the catastrophic loss of a given
data-centere.

[1]: http://blog.rightscale.com/2008/08/20/amazon-ebs-explained/

>>  * recovery from backup plans (we clearly have to automate (and test)
>> recovery off backups for this to be really useful)

OK, we should ticket this then.

Rufus

[...]