[PVE-User] Cluster HA withtou Fence Device !
admin-at-extremeshok-dot-com
admin at extremeshok.com
Mon Jul 7 12:20:13 CEST 2014
Assuming he is using shared storage (NAS)...
The nas operates as a 3rd node, and the systems are fenced via the NAS.
As long as 1 node can communicate with the NAS, it will have a higher
quorum, than the node that cant connect to the NAS.
Single point of failure is NAS, so its not HA.
On 7/7/2014 11:50 AM, Yannis Milios wrote:
>
> I tried the same config on a lab but does not work.
> I think its not possible to have HA without proper fencing.
> Im
>
> On Jul 7, 2014 12:40 PM, "Angel Docampo" <adocampo at dltec.net
> <mailto:adocampo at dltec.net>> wrote:
>
> What do you mean by "it works"?
> Without fence mechanism, the cluster won't be aware there a node
> down, and then you can continue writing modifications in both
> sides and finish with a split brain.
>
>
>
> El 05/07/14 21:49, Gilberto Nunes escribió:
>> WOW
>>
>> Even with just 2 nodes HA works fine!!!!
>>
>> I have tested some minutes ago...
>>
>> Is there something improvement on HA Code??
>>
>> I would like to know...
>>
>> Thanks
>>
>>
>> 2014-07-05 16:37 GMT-03:00 Gilberto Nunes
>> <gilberto.nunes32 at gmail.com <mailto:gilberto.nunes32 at gmail.com>>:
>>
>> Hi
>>
>> Today I have deploy a tree node plus storage scenario in a lab.
>> First at all, I image that I need configure fence device, as
>> docs demand...
>> So, I decide to use a MAnaged Switch (DLikn DES-3028).
>> But, I realize that even without fence configuratio HA works
>> properly and NICE, I might add..
>>
>> Here's the configuration:
>>
>> <?xml version="1.0"?>
>> <cluster config_version="18" name="CLUSTER">
>> <cman expected_votes="3"
>> keyfile="/var/lib/pve-cluster/corosync.authkey"/>
>> <quorumd allow_kill="0" interval="3" label="quorum"
>> tko="10" votes="1">
>> <heuristic interval="3" program="ping 10.90.90.100 -c1
>> -w1" score="1" tko="4"/>
>> <heuristic interval="3" program="ip addr | grep eth0 |
>> grep -q UP" score="2" tko="3"/>
>> </quorumd>
>> <totem token="54000"/>
>> <clusternodes>
>> <clusternode name="pve01" nodeid="1" votes="1">
>> </clusternode>
>> <clusternode name="pve02" nodeid="2" votes="1">
>> </clusternode>
>> <clusternode name="pve03" nodeid="3" votes="1">
>> </clusternode>
>> </clusternodes>
>> <rm>
>> <failoverdomains>
>> <failoverdomain name="serverfailover" ordered="1"
>> restricted="0">
>> <failoverdomainnode name="pve1" priority="1"/>
>> <failoverdomainnode name="pve2" priority="2"/>
>> <failoverdomainnode name="pve3" priority="3"/>
>> </failoverdomain>
>> </failoverdomains>
>> <pvevm autostart="1" vmid="100"/>
>> </rm>
>> </cluster>
>>
>>
>> I always though that I need fence device in order to make HA
>> work...
>>
>> Thanks
>>
>> --
>> Gilberto Ferreira
>>
>>
>>
>>
>> --
>> Gilberto Ferreira
>>
>>
>> _______________________________________________
>> pve-user mailing list
>> pve-user at pve.proxmox.com <mailto:pve-user at pve.proxmox.com>
>> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
>
> --
>
>
>
> *Angel Docampo
> *
> *Datalab Tecnologia, s.a.*
> Castillejos, 352 - 08025 Barcelona
> Tel. 93 476 69 14 - Ext: 706
> Mob. 670.299.381
>
>
> _______________________________________________
> pve-user mailing list
> pve-user at pve.proxmox.com <mailto:pve-user at pve.proxmox.com>
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
>
>
>
> _______________________________________________
> pve-user mailing list
> pve-user at pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.proxmox.com/pipermail/pve-user/attachments/20140707/87499e1d/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/png
Size: 5898 bytes
Desc: not available
URL: <http://lists.proxmox.com/pipermail/pve-user/attachments/20140707/87499e1d/attachment.png>
More information about the pve-user
mailing list