You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Jim Jagielski <ji...@jaguNET.com> on 2016/01/26 19:40:58 UTC

BalancerMembers (workers): stopped and disabled

Currently, the idea and "logic" associated w/ stopped and disabled
workers are kind of similar. There is a higher concept that one is
more 'admin' controlled and the other more 'autonomous' controlled,
but we really don't enforce any sort of conditions related to that.

I think it's time we start doing that, and use stopped and disabled
for similar but distinctly different conditions...

comments?

PS: yeah, this all comes about due to how the health-check module
    should interact w/ those states...

Re: BalancerMembers (workers): stopped and disabled

Posted by Jim Jagielski <ji...@jaguNET.com>.
Anymore thoughts about this? I'd like to start work on
this while I make some of the hcheck params changable via
the balancer-manager.

Re: BalancerMembers (workers): stopped and disabled

Posted by Jim Jagielski <ji...@jaguNET.com>.
> On Jan 26, 2016, at 2:27 PM, Ruediger Pluem <rp...@apache.org> wrote:
> 
> 
> 
> On 01/26/2016 08:11 PM, Jim Jagielski wrote:
>> My thoughts are that STOPPED means that health checks won't be
>> done, nor will retry be done. It means stopped-and-won't-automatically-
>> restart. Disabled is currently-offline-due-to-can't-be-accessed
>> and so health checks and retries will be done on those.
>> 
>> Or:
>> 
>> o Disabled: automatically detected as unavailable; will
>>             automatically re-enabled when the server is
>>             successfully retried/checked
> 
> How this this related to the error state of a worker?
> 

Basically, "disabled" would be "in error" as in, "I detected
an error on this worker" (I == httpd). stopped wouldn't be
considered "in error" because it wasn't remove due to
any sort of (automatic) error.

So stopped is admintratively removed; disabled is automatically removed,
as it were.

I'd need to check all the places where we check for a usable worker, a
worker in "error" etc, but that's all part of my plan to better define
the diffs between these 2 states.

> Regards
> 
> RĂ¼diger
> 
>> 
>> o Stopped:  Administratively stopped. Will never automatically
>>             be re-enabled; must be explicitly re-enabled.
>> 
>> 
>>> On Jan 26, 2016, at 1:58 PM, William A Rowe Jr <wr...@rowe-clan.net> wrote:
>>> 
>>> What semantics do you suggest for each?
>>> 
>>> True that they have been effectively identical (and redundant) so far...
>>> 
>>> On Jan 26, 2016 12:41, "Jim Jagielski" <ji...@jagunet.com> wrote:
>>> Currently, the idea and "logic" associated w/ stopped and disabled
>>> workers are kind of similar. There is a higher concept that one is
>>> more 'admin' controlled and the other more 'autonomous' controlled,
>>> but we really don't enforce any sort of conditions related to that.
>>> 
>>> I think it's time we start doing that, and use stopped and disabled
>>> for similar but distinctly different conditions...
>>> 
>>> comments?
>>> 
>>> PS: yeah, this all comes about due to how the health-check module
>>>    should interact w/ those states...
>> 
>> 


Re: BalancerMembers (workers): stopped and disabled

Posted by Ruediger Pluem <rp...@apache.org>.

On 01/26/2016 08:11 PM, Jim Jagielski wrote:
> My thoughts are that STOPPED means that health checks won't be
> done, nor will retry be done. It means stopped-and-won't-automatically-
> restart. Disabled is currently-offline-due-to-can't-be-accessed
> and so health checks and retries will be done on those.
> 
> Or:
> 
>  o Disabled: automatically detected as unavailable; will
>              automatically re-enabled when the server is
>              successfully retried/checked

How this this related to the error state of a worker?

Regards

RĂ¼diger

> 
>  o Stopped:  Administratively stopped. Will never automatically
>              be re-enabled; must be explicitly re-enabled.
> 
> 
>> On Jan 26, 2016, at 1:58 PM, William A Rowe Jr <wr...@rowe-clan.net> wrote:
>>
>> What semantics do you suggest for each?
>>
>> True that they have been effectively identical (and redundant) so far...
>>
>> On Jan 26, 2016 12:41, "Jim Jagielski" <ji...@jagunet.com> wrote:
>> Currently, the idea and "logic" associated w/ stopped and disabled
>> workers are kind of similar. There is a higher concept that one is
>> more 'admin' controlled and the other more 'autonomous' controlled,
>> but we really don't enforce any sort of conditions related to that.
>>
>> I think it's time we start doing that, and use stopped and disabled
>> for similar but distinctly different conditions...
>>
>> comments?
>>
>> PS: yeah, this all comes about due to how the health-check module
>>     should interact w/ those states...
> 
> 

Re: BalancerMembers (workers): stopped and disabled

Posted by Jim Jagielski <ji...@jaguNET.com>.
My thoughts are that STOPPED means that health checks won't be
done, nor will retry be done. It means stopped-and-won't-automatically-
restart. Disabled is currently-offline-due-to-can't-be-accessed
and so health checks and retries will be done on those.

Or:

 o Disabled: automatically detected as unavailable; will
             automatically re-enabled when the server is
             successfully retried/checked

 o Stopped:  Administratively stopped. Will never automatically
             be re-enabled; must be explicitly re-enabled.


> On Jan 26, 2016, at 1:58 PM, William A Rowe Jr <wr...@rowe-clan.net> wrote:
> 
> What semantics do you suggest for each?
> 
> True that they have been effectively identical (and redundant) so far...
> 
> On Jan 26, 2016 12:41, "Jim Jagielski" <ji...@jagunet.com> wrote:
> Currently, the idea and "logic" associated w/ stopped and disabled
> workers are kind of similar. There is a higher concept that one is
> more 'admin' controlled and the other more 'autonomous' controlled,
> but we really don't enforce any sort of conditions related to that.
> 
> I think it's time we start doing that, and use stopped and disabled
> for similar but distinctly different conditions...
> 
> comments?
> 
> PS: yeah, this all comes about due to how the health-check module
>     should interact w/ those states...


Re: BalancerMembers (workers): stopped and disabled

Posted by William A Rowe Jr <wr...@rowe-clan.net>.
What semantics do you suggest for each?

True that they have been effectively identical (and redundant) so far...
On Jan 26, 2016 12:41, "Jim Jagielski" <ji...@jagunet.com> wrote:

> Currently, the idea and "logic" associated w/ stopped and disabled
> workers are kind of similar. There is a higher concept that one is
> more 'admin' controlled and the other more 'autonomous' controlled,
> but we really don't enforce any sort of conditions related to that.
>
> I think it's time we start doing that, and use stopped and disabled
> for similar but distinctly different conditions...
>
> comments?
>
> PS: yeah, this all comes about due to how the health-check module
>     should interact w/ those states...
>