[Cluster-devel] logsys in fenced

[Fabio M. Di Nitto]

On Fri, 27 Jun 2008, David Teigland wrote:

> On Thu, Jun 26, 2008 at 04:43:28PM -0500, David Teigland wrote:
>> On Wed, Jun 25, 2008 at 04:21:55PM -0500, David Teigland wrote:
>>> I have yet to study logsys enough to propose an alternative to the macro
>>> magic that's now isolated at the end of fd.h.
>>
>> I've done some logsys studying and discussed it with Steve Dake.  The
>> logsys implementation and api is heavily geared toward use by openais
>> plugins (see the entire notion of "subsystems" that logsys is built
>> around).  Using it from a general program like fenced, not an
>> openais-plugin, makes the mismatch painfully clear.
>>
>> So, I don't hold out much hope that logsys is going to be useful for
>> non-plugins.  But, just because we don't share the actual code that
>> performs logging, doesn't mean we can't have a uniform approach for
>> setting up logging properies for all cluster-related programs.  That may
>> still be a worthwhile goal.
>
> The hope that I do still hold out for using logsys, is that we may be able
> to develop a simple syslog-like api for it, that would either parallel or
> sit above the plugin-oriented api that's there now.  Steve claims that the
> syslog(3) api is fundamentally broken and didn't sound at all interested
> in the idea, but maybe others are more open to the idea?

My goal is to have a common logging infrastructure with the same options 
all over and that will provide the same output etc. for virtually
everything as possible where it makes sense (daemons basically).

If you want to develop a parallel API or an API that sits on top, it's 
fine by me. Tho you will also have to take the responsibility to convert 
all code to use it, and update documentation and retain backward
compatibility for cman/qdisk/rgmanager that had already their own config 
sets prior logsys. I for sure don't want see again half code converted and 
half not.

In my head it sounds a lot of extra work to do but for sure i am not going 
to stop you, as long this is done before we declare our API stable for 3.0 
(that should happen within a reasonable amount of time == a month max).

Fabio

--
I'm going to make him an offer he can't refuse.