[Cluster-devel] [PATCH 1/3] Fix linking order

2012-11-02 Thread Jacek Konieczny
Objects should generally be listed before the libraries they require. DLM building fails for me without these changes. Signed-off-by: Jacek Konieczny --- dlm_controld/Makefile | 4 ++-- dlm_tool/Makefile | 2 +- fence/Makefile| 2 +- libdlm/Makefile | 8 4 files change

[Cluster-devel] [PATCH 3/3] Do not crash uncleanly on SIGINT and SIGHUP

2012-11-02 Thread Jacek Konieczny
Unclean shut-down of dlm_controld may trigger a node fencing, this should not be triggered by mistake. SIGINT is the natural way to stop dlm_controld started on a terminal with '-D' for debugging – let this work like SIGTERM. SIGHUP is often used for config reload. dlm_controld doesn't support th

[Cluster-devel] [PATCH 2/3] Run dlm.service after sys-kernel-config.mount

2012-11-02 Thread Jacek Konieczny
dlm_controld requires configfs mounted. systemd can take care for that, but it does it better if the order is set in the unit file. Signed-off-by: Jacek Konieczny --- init/dlm.service | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/init/dlm.service b/init/dlm.service index d7

[Cluster-devel] DLM patches from PLD Linux

2012-11-02 Thread Jacek Konieczny
Hello, These are three of the patches I have applied to dlm when packaging it for PLD Linux. I think they could be applied upstream. Greets, Jacek

Re: [Cluster-devel] [PATCH] gfs2-utils: build: Avoid using the kernel versions of kernel headers

2012-11-02 Thread Steven Whitehouse
Hi, Looks good to me, Steve. On Fri, 2012-11-02 at 13:57 +, Andrew Price wrote: > Currently the configure script checks for kernel headers at > /lib/modules/`uname -r`/source/include/linux/* which means we're using > kernel headers not meant to be used in userspace code. Thus we see a lot >

[Cluster-devel] [PATCH] gfs2-utils: build: Avoid using the kernel versions of kernel headers

2012-11-02 Thread Andrew Price
Currently the configure script checks for kernel headers at /lib/modules/`uname -r`/source/include/linux/* which means we're using kernel headers not meant to be used in userspace code. Thus we see a lot of "Attempt to use kernel headers from user space" warnings when building with a kernel-devel p