[sheepdog-users] Sheepdog 0.80: Queries and feedback

[Andrew J. Hobbs]

To shutdown the entire cluster at once:

dog cluster shutdown

To kill a specific node cleanly:

dog node kill <id>
    where id is the id from 'dog node list'

Don't kill a second node, or shutdown the cluster while a rebalance is in progress.

If you have multiple machines in separate racks, it could be useful to define each rack as a separate zone, but that would be for larger clusters where you're worried about a top-of-rack switch dying on you.


On 03/13/2014 04:12 AM, richter at ecos.de<mailto:richter at ecos.de> wrote:

5. The manual/help says in bold letters: "Remember: DO NOT shut down any of node before deactivating the cluster!

Hi, thank you for your feedbacks.
The second part of the sentence 'before deactivating the cluster' is to make sure there are no active recovery when you shut down the cluster.
I may have to test this because I think it shouldn't really cause problem but, if you want to be 100% sure of not getting issues when restarting the cluster, it may be better to stop it when it's not in recovery.

DO NOT shut down / kill a node after the other.". Yet isn't the point of sheepdog to be a distributed highly available storage system,
With that I meant "the right procedure to stop the cluster is not to shutdown/kill a node after the other".
I'm going to change a bit the doc so it's more clear.
[gr] So what is the correct way to shutdown a cluster? It’s still not clear to me
Thanks & Regards
Gerald




-------------- next part --------------
A non-text attachment was scrubbed...
Name: ajhobbs.vcf
Type: text/x-vcard
Size: 353 bytes
Desc: ajhobbs.vcf
URL: <http://lists.wpkg.org/pipermail/sheepdog-users/attachments/20140313/45cf5148/attachment-0005.vcf>