[Users] A question about Node RAM

[Tim Small]

On 06/01/12 22:59, jjs - mainphrame wrote:
> On Fri, Jan 6, 2012 at 12:34 PM, Tim Small <tim at seoss.co.uk
> <mailto:tim at seoss.co.uk>> wrote:
>
>
>     pacemaker+heartbeat
>
>
> Interesting idea, I wonder about the tradeoffs. I tend to keep the
> host node pretty lean and run heartbeat/corosync/pacemaker in the CTs,
> if anywhere.
>

We have a few machines where we put the OpenVZ container backing stores
on drbd and use heartbeat+pacemaker (we had some issues with corosync
during testing when we initially set things up a few years ago, but it's
probably fine now) to manage the OpenVZ containers as cluster resources.

Disk writes are relatively expensive so it's not perfect for all
workloads, but it works well overall, and has survived real hardware
failures (e.g. motherboard failure) with minimal downtime.

It also allows you to move nodes around easily and should allow you to
carry out things like host node kernel updates without bringing down
containers (using live migration to other HNs) - although we've not
gotten around to testing this.

Our machines are in pairs, but really it'd be better to have them in
something like groups of four, so that when a HN fails, the remaining 3
HNs each end up running a third of the evicted containers...  This would
require corosync instead of heartbeat of course (heartbeat supports 2
nodes only).

Tim.

-- 
South East Open Source Solutions Limited
Registered in England and Wales with company number 06134732.  
Registered Office: 2 Powell Gardens, Redhill, Surrey, RH1 1TQ
VAT number: 900 6633 53  http://seoss.co.uk/ +44-(0)1273-808309

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://openvz.org/pipermail/users/attachments/20120107/8fef3154/attachment.html