[Cluster-devel] [PATCH][RFC] add cman_qinit to libcman

[Cluster-devel] [PATCH][RFC] add cman_qinit to libcman

[Fabio Massimo Di Nitto]

Hi guys,

during the process of killing ccsd I noticed that almost all services that
communicate with cman wait to talk to cman or for the cluster to quorate
implementing over and over loops here and there.

The patch in attachment (basically untested) implements a simple cman_w(ait)init
call within libcman to take care of all of the above in one single call and it
allows to set the kind of requested socket (admin or not), a connection to cman
timeout, a quorum timeout and the standard privdata for cman_init and
cman_admin_init.

If this patch could go in HEAD, it will simplify a lot code within all the
subsystems that need cman to be up and running and possibly a quorated cluster
before they can work.

Please consider for inclusion.

Thanks
Fabio
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: cman_winit.diff
URL: <http://listman.redhat.com/archives/cluster-devel/attachments/20071105/f374fd51/attachment.ksh>

[Cluster-devel] [PATCH][RFC] add cman_winit to libcman

[Fabio Massimo Di Nitto]

Sorry for the previous mail.. it had the wrong patch in attachment.

Here is the right one.

Fabio

Fabio Massimo Di Nitto wrote:
> Hi guys,
> 
> during the process of killing ccsd I noticed that almost all services that
> communicate with cman wait to talk to cman or for the cluster to quorate
> implementing over and over loops here and there.
> 
> The patch in attachment (basically untested) implements a simple cman_w(ait)init
> call within libcman to take care of all of the above in one single call and it
> allows to set the kind of requested socket (admin or not), a connection to cman
> timeout, a quorum timeout and the standard privdata for cman_init and
> cman_admin_init.
> 
> If this patch could go in HEAD, it will simplify a lot code within all the
> subsystems that need cman to be up and running and possibly a quorated cluster
> before they can work.
> 
> Please consider for inclusion.
> 
> Thanks
> Fabio
> 

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: cman_winit.diff
URL: <http://listman.redhat.com/archives/cluster-devel/attachments/20071105/55dd2c04/attachment.ksh>

[Cluster-devel] [PATCH][RFC] add cman_winit to libcman

[Lon Hohberger]

On Mon, 2007-11-05 at 07:52 -0500, Fabio Massimo Di Nitto wrote:

ACK - but winit isn't working for me...

cman_wait_init ?
cman_init_wait ?

-- Lon

[Cluster-devel] [PATCH][RFC] add cman_winit to libcman

[Patrick Caulfield]

Lon Hohberger wrote:
> On Mon, 2007-11-05 at 07:52 -0500, Fabio Massimo Di Nitto wrote:
> 
> ACK - but winit isn't working for me...
> 
> cman_wait_init ?
> cman_init_wait ?
> 

I think it should be cman_wait_init()

In isolation I preferred cman_init_wait() but that looks inconsistent with thr
existing cman_admin_init().

with that change...

ACK
-- 
Patrick