[Cluster-devel] cluster/cman/daemon config.c

[Cluster-devel] cluster/cman/daemon config.c

[pcaulfield]

CVSROOT:	/cvs/cluster
Module name:	cluster
Changes by:	pcaulfield at sourceware.org	2006-08-15 13:43:12

Modified files:
	cman/daemon    : config.c 

Log message:
	"group" should be "amf"

Patches:
http://sourceware.org/cgi-bin/cvsweb.cgi/cluster/cman/daemon/config.c.diff?cvsroot=cluster&r1=1.17&r2=1.18

--- cluster/cman/daemon/config.c	2006/06/30 13:00:27	1.17
+++ cluster/cman/daemon/config.c	2006/08/15 13:43:12	1.18
@@ -131,7 +131,7 @@
 	read_config_for(cd, objdb, OBJECT_PARENT_HANDLE, "logging", "logging", 1);
 	read_config_for(cd, objdb, OBJECT_PARENT_HANDLE, "event", "event", 1);
 	read_config_for(cd, objdb, OBJECT_PARENT_HANDLE, "aisexec", "aisexec", 1);
-	read_config_for(cd, objdb, OBJECT_PARENT_HANDLE, "group", "group", 1);
+	read_config_for(cd, objdb, OBJECT_PARENT_HANDLE, "amf", "amf", 1);
 
 	/* This is stuff specific to us, eg quorum device timeout */
 	read_config_for(cd, objdb, OBJECT_PARENT_HANDLE, "cman", "cman", 1);

[Cluster-devel] cluster/cman/daemon config.c

[ccaulfield]

CVSROOT:	/cvs/cluster
Module name:	cluster
Changes by:	ccaulfield at sourceware.org	2008-02-18 17:25:42

Modified files:
	cman/daemon    : config.c 

Log message:
	Allow unnamed parent objects. This fixes a bug where
	<fence> entries appeared under the top-level rather than the clusternode.

Patches:
http://sourceware.org/cgi-bin/cvsweb.cgi/cluster/cman/daemon/config.c.diff?cvsroot=cluster&r1=1.20&r2=1.21

--- cluster/cman/daemon/config.c	2008/01/02 16:35:44	1.20
+++ cluster/cman/daemon/config.c	2008/02/18 17:25:41	1.21
@@ -115,6 +115,10 @@
 		{
 			char subpath[1024];
 
+			/* Allow for empty parents */
+			if (!object_handle)
+				object_handle = parent;
+
 			/* Found a subkey, iterate through it's sub sections */
 			sprintf(subpath, "%s/%s[%d]", key, str, ++count);
 			if (!read_config_for(ccs_fd, objdb, object_handle, str, subpath, 0))

[Cluster-devel] cluster/cman/daemon config.c

[Fabio M. Di Nitto]

Was this commit really ment to go to CVS?

Fabio

On Mon, 18 Feb 2008, ccaulfield at sourceware.org wrote:

> CVSROOT:	/cvs/cluster
> Module name:	cluster
> Changes by:	ccaulfield at sourceware.org	2008-02-18 17:25:42
>
> Modified files:
> 	cman/daemon    : config.c
>
> Log message:
> 	Allow unnamed parent objects. This fixes a bug where
> 	<fence> entries appeared under the top-level rather than the clusternode.
>
> Patches:
> http://sourceware.org/cgi-bin/cvsweb.cgi/cluster/cman/daemon/config.c.diff?cvsroot=cluster&r1=1.20&r2=1.21
>
> --- cluster/cman/daemon/config.c	2008/01/02 16:35:44	1.20
> +++ cluster/cman/daemon/config.c	2008/02/18 17:25:41	1.21
> @@ -115,6 +115,10 @@
> 		{
> 			char subpath[1024];
>
> +			/* Allow for empty parents */
> +			if (!object_handle)
> +				object_handle = parent;
> +
> 			/* Found a subkey, iterate through it's sub sections */
> 			sprintf(subpath, "%s/%s[%d]", key, str, ++count);
> 			if (!read_config_for(ccs_fd, objdb, object_handle, str, subpath, 0))
>


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