Xen users - Jul 2006 - xen-3.0.2-2 and redhat-cluster (gfs/clvm/etc)

Hi all,

i was just wondering if anyone has successfully compiled redhat-cluster with a 
2.6.16 xen0 kernel?

Have also posted to redhat-cluster mailinglist, but though i could try here??

I have tried compiling (on debian sarge) with the following redhat-cluster 
sources
ftp://sources.redhat.com/pub/cluster/releases/cluster-1.01.00.tar.gz
ftp://sources.redhat.com/pub/cluster/releases/cluster-1.02.00.tar.gz
and 
cvs -d :pserver:cvs@sources.redhat.com:/cvs/cluster checkout -r STABLE cluster

but get various compilation errors with all

All build attempts have been built against the kernel tree in 
 /xen-3.0.2-2/linux-2.6.16-xen0

Here is the error for cluster-1.01
/usr/src/cluster-1.01.00/gfs-kernel/src/gfs/ops_file.c: In function 
`gfs_write'':
/usr/src/cluster-1.01.00/gfs-kernel/src/gfs/ops_file.c:934: error: structure 
has no member named `i_sem''
/usr/src/cluster-1.01.00/gfs-kernel/src/gfs/ops_file.c:939: error: structure 
has no member named `i_sem''
make[4]: *** [/usr/src/cluster-1.01.00/gfs-kernel/src/gfs/ops_file.o] Error 1
make[3]: *** [_module_/usr/src/cluster-1.01.00/gfs-kernel/src/gfs] Error 2
make[3]: Leaving directory
`/home/jfrazer/xen-3.0.2-2/linux-2.6.16-xen0''
make[2]: *** [all] Error 2
make[2]: Leaving directory
`/usr/src/cluster-1.01.00/gfs-kernel/src/gfs''
make[1]: *** [install] Error 2
make[1]: Leaving directory `/usr/src/cluster-1.01.00/gfs-kernel/src''
make: *** [install] Error 2


and for cluster-1.02
make[5]: Entering directory
`/usr/src/cluster-1.02.00/gnbd/tools/gnbd_export''
gcc -Wall -I../../include -I../../server -I../../utils -I../../config
-I//usr/include -O2 -DGNBD_RELEASE_NAME=\"1.02.00\"
gnbd_export.c ../../utils/gnbd_utils.c ../../utils/trans.c  -L//usr/lib -lmagma
-ldl -lpthread -lsysfs -o
gnbd_export
gnbd_export.c: In function `get_sysfs_name'':
gnbd_export.c:358: warning: implicit declaration of function 
`sysfs_open_directory''
gnbd_export.c:358: warning: assignment makes pointer from integer without a 
cast
gnbd_export.c:364: warning: implicit declaration of function 
`sysfs_get_dir_subdirs''
gnbd_export.c:364: warning: assignment makes pointer from integer without a 
cast
gnbd_export.c:372: warning: implicit declaration of function 
`sysfs_get_directory_attribute''
gnbd_export.c:372: warning: assignment makes pointer from integer without a 
cast
gnbd_export.c:384: error: dereferencing pointer to incomplete type
gnbd_export.c:384: error: dereferencing pointer to incomplete type
gnbd_export.c:384: error: dereferencing pointer to incomplete type
gnbd_export.c:384: error: dereferencing pointer to incomplete type
gnbd_export.c:384: error: dereferencing pointer to incomplete type
gnbd_export.c:384: error: dereferencing pointer to incomplete type
gnbd_export.c:384: error: dereferencing pointer to incomplete type
gnbd_export.c:392: warning: implicit declaration of function 
`sysfs_close_directory''
make[5]: *** [gnbd_export] Error 1
make[5]: Leaving directory
`/usr/src/cluster-1.02.00/gnbd/tools/gnbd_export''
make[4]: *** [copytobin] Error 2
make[4]: Leaving directory `/usr/src/cluster-1.02.00/gnbd/tools''
make[3]: *** [all] Error 2
make[3]: Leaving directory `/usr/src/cluster-1.02.00/gnbd''
make[2]: *** [all] Error 2
make[2]: Leaving directory `/usr/src/cluster-1.02.00/gnbd/bin''
make[1]: *** [install] Error 2
make[1]: Leaving directory `/usr/src/cluster-1.02.00/gnbd''
make: *** [install] Error 2



and for cvs
rm -f linux lm_interface.h
ln -s . linux
ln -s ../../src/harness/lm_interface.h .
make -C /home/jfrazer/xen-3.0.2-2/linux-2.6.16-xen0 
M=/usr/src/cluster/gfs-kernel/src/nolock 
symverfile=/usr/src/cluster/gfs-kernel/src/nolock/../harness/lock_harness.symvers
modules USING_KBUILD=yes
make[4]: Entering directory
`/home/jfrazer/xen-3.0.2-2/linux-2.6.16-xen0''
  CC [M]  /usr/src/cluster/gfs-kernel/src/nolock/main.o
/usr/src/cluster/gfs-kernel/src/nolock/main.c: In function
`nolock_plock_get'':
/usr/src/cluster/gfs-kernel/src/nolock/main.c:250: error: too many arguments 
to function `posix_test_lock''
/usr/src/cluster/gfs-kernel/src/nolock/main.c:250: warning: assignment makes 
integer from pointer without a cast
make[5]: *** [/usr/src/cluster/gfs-kernel/src/nolock/main.o] Error 1
make[4]: *** [_module_/usr/src/cluster/gfs-kernel/src/nolock] Error 2
make[4]: Leaving directory
`/home/jfrazer/xen-3.0.2-2/linux-2.6.16-xen0''
make[3]: *** [all] Error 2
make[3]: Leaving directory `/usr/src/cluster/gfs-kernel/src/nolock''
make[2]: *** [install] Error 2
make[2]: Leaving directory `/usr/src/cluster/gfs-kernel/src''
make[1]: *** [install] Error 2
make[1]: Leaving directory `/usr/src/cluster/gfs-kernel''
make: *** [install] Error 2

If anyone has any suggestions or can tell me what sources from redhat-cluster 
will build against which kernels would be great!!!

Many thanks

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Hi,

On 7/24/06, Jim Frazer <jim-bo@hotpop.com> wrote:
> [...]
> All build attempts have been built against the kernel tree in
>  /xen-3.0.2-2/linux-2.6.16-xen0
Maybe you try it with xen testing or unstable?
> [...] (100+ lines of compile errors omitted)
> make[1]: Leaving directory
`/usr/src/cluster-1.01.00/gfs-kernel/src''
> make: *** [install] Error 2
> [...]
Please don''t post such long error logs without being requested to do
so.
Somtimes it helps to post some log output here with a question, but
not in the quantities as it seems to become standard here - everybody
is just copying their full logs into a mail.

If nobody has an idea of this cluster stuff, these logs won''t help. If
there''s somebody who knows about it, he might be able to help without
logs, if thinks he can help, but needs an error log to know exactly
the problem, he will ask for it. These logs make it less likely to get
responses, because nobody has time to read 2 pages of compile errors.
> If anyone has any suggestions or can tell me what sources from
redhat-cluster
> will build against which kernels would be great!!!
I don''t think the xen community will be able to tell you for which
kernel version the redhat patches are made - so better ask the
developers of these patches - if they don''t know themselves or
don''t
reply, then this is probably not the kind of software anybody wants to
use.
Maybe you are unlucky and they will only work nicely with a kernel
full of other redhat-specific patches.

Henning

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Henning Sprang wrote:
> Hi,
> 
> On 7/24/06, Jim Frazer <jim-bo@hotpop.com> wrote:
>> [...]
>> All build attempts have been built against the kernel tree in
>>  /xen-3.0.2-2/linux-2.6.16-xen0
> 
> Maybe you try it with xen testing or unstable?
> 
Cluster Suite 1.02.00 (withe some important patches, also in Portage)
builds fine for Gentoo Portage for 2.6.16.26 patched with Xen 3.0.2.
Now, if only the VMs would stop stalling and getting fenced all of the
time...

-- 
Christopher G. Stach II

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> Cluster Suite 1.02.00 (withe some important patches, also in Portage)
> builds fine for Gentoo Portage for 2.6.16.26 patched with Xen 3.0.2.
> Now, if only the VMs would stop stalling and getting fenced all of the
> time...
Does this mean you are able to fail over virtual machines?





_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Matthijs ter Woord wrote:
>> Cluster Suite 1.02.00 (withe some important patches, also in Portage)
>> builds fine for Gentoo Portage for 2.6.16.26 patched with Xen 3.0.2.
>> Now, if only the VMs would stop stalling and getting fenced all of the
>> time...
> 
> Does this mean you are able to fail over virtual machines?
In a LVS sort of way?  Yes, although failover in a cluster that is using
 GFS only works as well as automated or manual fencing does.  Since I
can''t get my domU nodes to ssh into the dom0 for fence_xen to work (Xen
networking issue of some sort that I don''t have time to figure out), I
have to use manual xm destroy and fence_ack_manual.  It''s a real pain
in
the ass.  It wouldn''t be an issue at all if the domUs didn''t
miss so
many heartbeats.  I have a hunch that the credit scheduler in 3.0.3 is
going to help that a lot.

-- 
Christopher G. Stach II

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

> Please don''t post such long error logs without being requested to
do so.
no worries...sorry about that...I was thinking it may be more helpful than 
painful to include them...wrong I was 
> I don''t think the xen community will be able to tell you for which
> kernel version the redhat patches are made
understood...no problem...i was only asking because i''ve seen a largish
thread
on xen list regarding iscsi, clvm, drbd and the like (clvm is what i''m
after)
and thought that people may be able to shed some light

In any case, sorry to annoy, and to follow up, thanks Christopher for gentoo 
suggestion. Built fine on gentoo.

Ended up compiling 1.02 fine on debian with 3.0.2-2 xen after figuring out 
some libraries it needed and doing a fresh build of the xen kernel tree
For anyone who may read this having similar issues these packages tended to 
help things along
libsysfs-dev
ncurses-dev
libxml2-dev

Cheers







_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Jim wrote:
> Ended up compiling 1.02 fine on debian with 3.0.2-2 xen after figuring out 
> some libraries it needed and doing a fresh build of the xen kernel tree
> For anyone who may read this having similar issues these packages tended to
> help things along
> libsysfs-dev
> ncurses-dev
> libxml2-dev
Be sure to get those patches against RH''s CVS that are in
Gentoo''s
Portage, else you''ll probably hate life when every cluster node crashes
when a node leaves properly. :)

-- 
Christopher G. Stach II

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users