xen kernel 2.6.16.60-0.27 vs HA 2.1.3

[Paul.Thurn]

Hi there!

Got a setup with two nodes SLES10SP2, xen kernel 2.6.16.60-0.27-xen.
Also HA 2.1.3, which didn't work since the SP1 to SP2 upgrade, had a case at
novell, and is now freshly installed and is working. there was a previous
fencing problem.

Resources setup is imagestorecloneset and configstorecloneset which works,
and some vm for testing purpose, which don't work.

The server are connected to a san, so that all servers (later five of them) are
seeing all xen resources and images, at the moment there are just to hkx1fs and
hkx2fs.

Xen is working, starting and stopping servers from gui or commandline is no
problem.
Starting the vm from ha commandline works eg:
  export OCF_ROOT='/usr/lib/ocf'
  export OCF_RESKEY_xmfile='/etc/xen/vm/sles10test'
  export OCF_RESKEY_allow_migrate=1

   cd /usr/lib/ocf/resource.d/heartbeat
  ./Xen start

But starting the vm from HA gui doesn't work.

cibadmin -Q is availible on ftp.hkr.se under /HA as cibQ and
ha-debug is also there.
login: anonymous
I think it's to big to be in a mail.

i would appreciate any support on this matter, and if the are any tests or
other logfiles needed, I would gladly do that for you.

Thanks in advance /Paul



------------------------------------------------------------------
 Paul Thurn,         291 88 Kristianstad    
 Tel. [phone]         Fax. [phone]    
 Email: Paul.Thurn@...        
------------------------------------------------------------------
Visit the Scandinavian Pegasus site. http://www.pegasus.hkr.se
------------------------------------------------------------------
Never argue with an idiot. They drag you down to their level and beat you with
experience.
------------------------------------------------------------------

_______________________________________________
Linux-HA mailing list
Linux-HA@...
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

[Lars Marowsky-Bree]

On 2008-09-29T11:00:04, "Paul.Thurn" <Paul_T@...> wrote:

> But starting the vm from HA gui doesn't work.

Doesn't work "how"?

BTW:

<clone id="configstorecloneset" notify="true" globally_unique="true">
<clone id="imagestorecloneset" notify="true" globally_unique="true">

globally_unique must be set to false here (this might require a full
restart of the cluster). I'm surprised this works at all.

For the Xen RA:

       <primitive id="s10test" class="ocf" type="Xen" provider="heartbeat">
         <meta_attributes id="s10test_meta_attrs">
           <attributes>
             <nvpair id="s10test_metaattr_target_role" name="target_role" value="stopped"/>
           </attributes>
         </meta_attributes>
         <instance_attributes id="s10test_instance_attrs">
           <attributes>
             <nvpair id="4279698e-95c9-4d94-8d2f-73345566be0c" name="xmfile" value="/etc/xen/vm/sles10test.xml"/>
             <nvpair id="d040f12e-bf93-4d58-9fa7-2333763c1865" name="allow_migrate" value="1"/>
           </attributes>
         </instance_attributes>
         <operations>
           <op id="81762d05-cdfb-40e6-bd3b-37c5579a4bbd" name="start" description="strat" timeout="60" start_delay="0" disabled="false" role="Started" interval="10"/>
           <op id="5b13b767-d362-4df1-be1b-69ae4eb95a25" name="stop" description="stoopy" timeout="40" start_delay="30" disabled="false" role="Started" interval="10"/>
           <op id="9b657f82-5bf0-452b-b82d-e1eeb6e6ed33" name="status" description="ststs" interval="10" timeout="30" start_delay="120" disabled="false" role="Started"/>
           <op id="4b65aac6-d2be-4e75-a303-1f1a96f68d84" name="meta-data" description="meta" interval="5" timeout="5" start_delay="0" disabled="false" role="Started"/>
           <op id="07849bbd-6326-4698-a42e-62fa2f928884" name="monitor" description="moni" interval="10" timeout="30" start_delay="120" disabled="false" role="Started"/>
         </operations>
       </primitive>


allow_migrate should be a meta_attribute.

And there's no point to define a "status" operation here. That needs to
be removed.

The status section indicates that start failed. Read the logs as to
why. But since you don't have rules ordering and collocating the Xen
start after the filesystem, that would be the reason ...


Regards,
    Lars

--
Teamlead Kernel, SuSE Labs, Research and Development
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
"Experience is the name everyone gives to their mistakes." -- Oscar Wilde

_______________________________________________
Linux-HA mailing list
Linux-HA@...
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

[Paul.Thurn]

> On 2008-09-29T11:00:04, "Paul.Thurn" <Paul_T@...> wrote:
>
> > But starting the vm from HA gui doesn't work.
>
> Doesn't work "how"?
>
the xen vm  (in this case s10test) doesn't start.
crm_mon  indicate timed out. and looking in xm list on
either server doesn't indicate the s10test has been started.

> BTW:
>
> <clone id="configstorecloneset" notify="true" globally_unique="true">
> <clone id="imagestorecloneset" notify="true" globally_unique="true">
>
> globally_unique must be set to false here (this might require a full
> restart of the cluster). I'm surprised this works at all.
>
globally_unique changed to false, all heartbeat cluster nodes shutdown,
and then all cluster nodes started again.

> For the Xen RA:
>
>        <primitive id="s10test" class="ocf" type="Xen" provider="heartbeat">
>          <meta_attributes id="s10test_meta_attrs">
>            <attributes>
>              <nvpair id="s10test_metaattr_target_role" name="target_role" value="stopped"/>
>            </attributes>
>          </meta_attributes>
>          <instance_attributes id="s10test_instance_attrs">
>            <attributes>
>              <nvpair id="4279698e-95c9-4d94-8d2f-73345566be0c" name="xmfile" value="/etc/xen/vm/sles10test.xml"/>
>              <nvpair id="d040f12e-bf93-4d58-9fa7-2333763c1865" name="allow_migrate" value="1"/>
>            </attributes>
>          </instance_attributes>
>          <operations>
>            <op id="81762d05-cdfb-40e6-bd3b-37c5579a4bbd" name="start" description="strat" timeout="60" start_delay="0" disabled="false" role="Started" interval="10"/>
>            <op id="5b13b767-d362-4df1-be1b-69ae4eb95a25" name="stop" description="stoopy" timeout="40" start_delay="30" disabled="false" role="Started" interval="10"/>
>            <op id="9b657f82-5bf0-452b-b82d-e1eeb6e6ed33" name="status" description="ststs" interval="10" timeout="30" start_delay="120" disabled="false" role="Started"/>
>            <op id="4b65aac6-d2be-4e75-a303-1f1a96f68d84" name="meta-data" description="meta" interval="5" timeout="5" start_delay="0" disabled="false" role="Started"/>
>            <op id="07849bbd-6326-4698-a42e-62fa2f928884" name="monitor" description="moni" interval="10" timeout="30" start_delay="120" disabled="false" role="Started"/>
>          </operations>
>        </primitive>
>
>
> allow_migrate should be a meta_attribute.
>

Through the gui, allow_migrate lies under parameters, that's were i added it.
i could change it through a commandline import, should i do that?

> And there's no point to define a "status" operation here. That needs to
> be removed.
>
Removed.

> The status section indicates that start failed. Read the logs as to
> why. But since you don't have rules ordering and collocating the Xen
> start after the filesystem, that would be the reason ...
>
hmm I'm lost here.. what am i supposed to do/test?

>
> Regards,
>     Lars
>

/paul

------------------------------------------------------------------
 Paul Thurn,         291 88 Kristianstad    
 Tel. [phone]         Fax. [phone]    
 Email: Paul.Thurn@...        
------------------------------------------------------------------
Visit the Scandinavian Pegasus site. http://www.pegasus.hkr.se
------------------------------------------------------------------
Never argue with an idiot. They drag you down to their level and beat you with
experience.
------------------------------------------------------------------

_______________________________________________
Linux-HA mailing list
Linux-HA@...
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

[Lars Marowsky-Bree]

On 2008-09-29T14:22:49, "Paul.Thurn" <Paul_T@...> wrote:

> > The status section indicates that start failed. Read the logs as to
> > why. But since you don't have rules ordering and collocating the Xen
> > start after the filesystem, that would be the reason ...
> hmm I'm lost here.. what am i supposed to do/test?

You need to add rsc_colocation + rsc_order constraints.


Regards,
    Lars

--
Teamlead Kernel, SuSE Labs, Research and Development
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
"Experience is the name everyone gives to their mistakes." -- Oscar Wilde

_______________________________________________
Linux-HA mailing list
Linux-HA@...
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

[Paul.Thurn]

> On 2008-09-29T14:22:49, "Paul.Thurn" <Paul_T@...> wrote:
>
> > > The status section indicates that start failed. Read the logs as to
> > > why. But since you don't have rules ordering and collocating the Xen
> > > start after the filesystem, that would be the reason ...
> > hmm I'm lost here.. what am i supposed to do/test?
>
> You need to add rsc_colocation + rsc_order constraints.
>
I'm sorry.... i'm just a newbie on this......(old netware dude)
But if i just want a xen virtual machine to start on any of the cluster nodes.
do I really need colocation and order. I understand that i need it in a later
stage, when everything are going to work automatically after a restart/boot.

If Location depends on colocation and order I will remove location, it was just
a suggestion from novells sles.virtualization forum, to try if it would start
on hkx1fs with the heartbeat unloaded on hkx2fs.

The most easiest way for me to add order and colocation is through the
gui (I'm not a script kiddie yet), I've studied the examples on linux-ha.org,
but i just can't work it out - sorry. So in case of the requirement of order
and colocation, i would really appriciate further instructions.

/paul

>
> Regards,
>     Lars
>
> --
> Teamlead Kernel, SuSE Labs, Research and Development
> SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
> "Experience is the name everyone gives to their mistakes." -- Oscar Wilde
>
> _______________________________________________
> Linux-HA mailing list
> Linux-HA@...
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>
> --
> This message has been scanned for viruses and
> dangerous content by ESVA, and is
> believed to be clean.
>

------------------------------------------------------------------
 Paul Thurn,         291 88 Kristianstad
 Tel. [phone]         Fax. [phone]
 Email: Paul.Thurn@...
------------------------------------------------------------------
Visit the Scandinavian Pegasus site. http://www.pegasus.hkr.se
------------------------------------------------------------------
Never argue with an idiot. They drag you down to their level and beat you with
experience.
------------------------------------------------------------------

_______________________________________________
Linux-HA mailing list
Linux-HA@...
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

[Dejan Muhamedagic]

Hi,

On Tue, Sep 30, 2008 at 08:45:14AM +0100, Paul.Thurn wrote:

> > On 2008-09-29T14:22:49, "Paul.Thurn" <Paul_T@...> wrote:
> >
> > > > The status section indicates that start failed. Read the logs as to
> > > > why. But since you don't have rules ordering and collocating the Xen
> > > > start after the filesystem, that would be the reason ...
> > > hmm I'm lost here.. what am i supposed to do/test?
> >
> > You need to add rsc_colocation + rsc_order constraints.
> >
> I'm sorry.... i'm just a newbie on this......(old netware dude)
> But if i just want a xen virtual machine to start on any of the cluster nodes.
> do I really need colocation and order. I understand that i need it in a later
> stage, when everything are going to work automatically after a restart/boot.

In this case you need it because the xen resource depends on the
filesystem.

> If Location depends on colocation and order I will remove location, it was just
> a suggestion from novells sles.virtualization forum, to try if it would start
> on hkx1fs with the heartbeat unloaded on hkx2fs.

A colocation and order are usually mandatory, whereas a location
typically indicates a preference.

> The most easiest way for me to add order and colocation is through the
> gui (I'm not a script kiddie yet), I've studied the examples on linux-ha.org,
> but i just can't work it out - sorry. So in case of the requirement of order
> and colocation, i would really appriciate further instructions.

Sort of hard to give gui instructions, at least for me. I think
that there are a few tutorials at linux-ha.org of which some, I
guess, include the gui way.

BTW, you don't need to be a "script kiddie", it's all XML which
is an entirely different matter ;-)

Thanks,

Dejan

> /paul
>
> >
> > Regards,
> >     Lars
> >
> > --
> > Teamlead Kernel, SuSE Labs, Research and Development
> > SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG N?rnberg)
> > "Experience is the name everyone gives to their mistakes." -- Oscar Wilde
> >
> > _______________________________________________
> > Linux-HA mailing list
> > Linux-HA@...
> > http://lists.linux-ha.org/mailman/listinfo/linux-ha
> > See also: http://linux-ha.org/ReportingProblems
> >
> > --
> > This message has been scanned for viruses and
> > dangerous content by ESVA, and is
> > believed to be clean.
> >
>
>
> ------------------------------------------------------------------
>  Paul Thurn,         291 88 Kristianstad
>  Tel. [phone]         Fax. [phone]
>  Email: Paul.Thurn@...
> ------------------------------------------------------------------
> Visit the Scandinavian Pegasus site. http://www.pegasus.hkr.se
> ------------------------------------------------------------------
> Never argue with an idiot. They drag you down to their level and beat you with
> experience.
> ------------------------------------------------------------------
>
> _______________________________________________
> Linux-HA mailing list
> Linux-HA@...
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems

_______________________________________________
Linux-HA mailing list
Linux-HA@...
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

[Paul.Thurn]

Hi, thank you for our answer...

> Hi,
>
> On Tue, Sep 30, 2008 at 08:45:14AM +0100, Paul.Thurn wrote:
> > > On 2008-09-29T14:22:49, "Paul.Thurn" <Paul_T@...> wrote:
> > >
> > > > > The status section indicates that start failed. Read the logs as to
> > > > > why. But since you don't have rules ordering and collocating the Xen
> > > > > start after the filesystem, that would be the reason ...
> > > > hmm I'm lost here.. what am i supposed to do/test?
> > >
> > > You need to add rsc_colocation + rsc_order constraints.
> > >
> > I'm sorry.... i'm just a newbie on this......(old netware dude)
> > But if i just want a xen virtual machine to start on any of the cluster nodes.
> > do I really need colocation and order. I understand that i need it in a later
> > stage, when everything are going to work automatically after a restart/boot.
>
> In this case you need it because the xen resource depends on the
> filesystem.

we are starting the xen resource manually.... and as you point out, the filesys
got to be online.
But this is just to see if we can get it to work at all. We had it working at
sles10sp1, but after the sp2 we can't get any xen resources to work.

>
> > If Location depends on colocation and order I will remove location, it was just
> > a suggestion from novells sles.virtualization forum, to try if it would start
> > on hkx1fs with the heartbeat unloaded on hkx2fs.
>
> A colocation and order are usually mandatory, whereas a location
> typically indicates a preference.
>
> > The most easiest way for me to add order and colocation is through the
> > gui (I'm not a script kiddie yet), I've studied the examples on linux-ha.org,
> > but i just can't work it out - sorry. So in case of the requirement of order
> > and colocation, i would really appriciate further instructions.
>
> Sort of hard to give gui instructions, at least for me. I think
> that there are a few tutorials at linux-ha.org of which some, I
> guess, include the gui way.
>
> BTW, you don't need to be a "script kiddie", it's all XML which
> is an entirely different matter ;-)
>
> Thanks,
>
> Dejan
>

All the xml is also a new world for me, takes some time to adapt it.
And using examples from linux-ha.org isn't always good, seems there are
a misch-masch of all kind of versions.
The next step we are taking is uninstall all heartbeat, and erase all files
that has to do with heartbeat, then make a new install 2.1.4 and se were it
leads.

/paul


------------------------------------------------------------------
 Paul Thurn,         291 88 Kristianstad    
 Tel. [phone]         Fax. [phone]    
 Email: Paul.Thurn@...        
------------------------------------------------------------------
Visit the Scandinavian Pegasus site. http://www.pegasus.hkr.se
------------------------------------------------------------------
Never argue with an idiot. They drag you down to their level and beat you with
experience.
------------------------------------------------------------------

_______________________________________________
Linux-HA mailing list
Linux-HA@...
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

[Paul.Thurn]

Got the whole thing rolling......

Uninstalled the HA
Erased all directories that had anything to do with HA
did a file scan on the server and found, some more files...

/sys/kernel/config/cluster/ocfs2
/sys/o2cb/logmask

that had with heartbeat to do.

uninstalled ocfs2

installed ocfs2 again, with a brand new configuration.
found errors with an old cluster node x3.
restarted x3, no more errors.

installed HA, added the nodes and resources..

tada.... works!
without order and colocation constraints, these will be put on later.

So if the problem lied within the ocfs2 or the heartbeat is difficult to say,
but i'm just glad it works.

thank you all for your engagement.

/paul

------------------------------------------------------------------
 Paul Thurn,         291 88 Kristianstad    
 Tel. [phone]         Fax. [phone]    
 Email: Paul.Thurn@...        
------------------------------------------------------------------
Visit the Scandinavian Pegasus site. http://www.pegasus.hkr.se
------------------------------------------------------------------
Never argue with an idiot. They drag you down to their level and beat you with
experience.
------------------------------------------------------------------

_______________________________________________
Linux-HA mailing list
Linux-HA@...
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems