On Tue, Feb 15, 2011 at 12:20 PM, Iustin Pop <[email protected]> wrote:
> On Tue, Feb 15, 2011 at 11:51:15AM +0100, René Nussbaumer wrote:
>> This command came up in a local discussion about how much logic we
>> should put into gnt-node power off. We decided to go with a separate
>> convenience command which takes care of all the steps through the
>> procedures to turn down the cluster into a state suitable for a shutdown
>> of the underlying machine.
>> ---
>>  doc/design-oob.rst |   23 +++++++++++++++++++++++
>>  1 files changed, 23 insertions(+), 0 deletions(-)
>>
>> diff --git a/doc/design-oob.rst b/doc/design-oob.rst
>> index 1e78c1a..43ef40f 100644
>> --- a/doc/design-oob.rst
>> +++ b/doc/design-oob.rst
>> @@ -52,6 +52,29 @@ New ``gnt-cluster`` Parameter
>>  | Parameters: ``--oob-program``
>>  | Options: ``--oob-program``: executable OOB program (absolute path)
>>
>> +New ``gnt-cluster epo`` Command
>> ++++++++++++++++++++++++++++++++
>> +
>> +| Program: ``gnt-cluster``
>> +| Command: ``epo``
>> +| Parameter: ``--on``
>> +| Options: ``--on``: By default epo turns off, with ``--on`` it tries to 
>> get the
>> +|                    cluster back online
>
> Add --confirm-shutdown so that the command can be run non-interactively.

To be consistent with the rest --force that would be, no?

>> +
>> +This is a convenience command to allow easy emergency power off of a whole
>> +cluster or part of it. It takes care of all steps needed to get the cluster 
>> into
>> +a sane state to turn off the nodes.
>> +
>> +With ``--on`` it does the reverse and tries to bring the rest of the 
>> cluster back
>> +to life.
>> +
>> +.. note::
>> +  The master node is not able to shut itself cleanly down. Therefore, this
>> +  command will not do all the work on single node clusters. On multi node
>> +  clusters the command tries to find another master or if that is not 
>> possible
>> +  prepares everything to the point where the user has to shutdown the master
>> +  node itself alone this applies also to the single node cluster 
>> configuration.
>
> Sorry, what do you mean, "tried to find another master"? Do you mean
> doing a master failover if we try to shutdown current node group?

Yes, if the master happens to be in the selected nodes (though node
groups or manually specified) it will do a master-failover to another
available node in the cluster which isn't affected by the shutdown.

Is this functionality appreciated?

René

Reply via email to