Cleveland, OH 44115
>
> Direct: 216-373-4808
> Fax: 216-373-4657
> ega...@broadvox.net
>
>
> CONFIDENTIAL COMMUNICATION. This e-mail and any files transmitted with it
> are confidential and are intended solely for the use of the individual or
> entity to whom
iot Gable
> Senior Engineer
> 1228 Euclid Ave, Suite 390
> Cleveland, OH 44115
>
> Direct: 216-373-4808
> Fax: 216-373-4657
> ega...@broadvox.net
>
>
> CONFIDENTIAL COMMUNICATION. This e-mail and any files transmitted with it
> are confidential and are in
657
> ega...@broadvox.net
>
>
> CONFIDENTIAL COMMUNICATION. This e-mail and any files transmitted with it
> are confidential and are intended solely for the use of the individual or
> entity to whom it is addressed. If you are not the intended recipient, please
> call me
If you use scripted input to crm, use full command lines:
CRM < In the past, I have always done things by manually creating a CIB XML
> file and then importing it. But, to save time, I thought I would try
> CRM. So, I made this script:
>
>
>
> #!/bin/bash
>
>
>
> crm<
> configure
>
> e
On Thu, 2009-10-29 at 10:00 -0400, Eliot Gable wrote:
> Once I modify configure.ac to remove this flag and regenerate
> everything,
you could have used:
./configure --disable-fatal-warnings
> I run into linking errors because the Makefile LIBS= line tells
> Pacemaker to use /usr/lib and /lib64
On Thu, 2009-10-22 at 15:10 +0200, Florian Haas wrote:
> On 10/22/2009 02:37 PM, Andrew Beekhof wrote:
> >> I wondered, does it happen dynamically? If one resource starts using a
> >> lot of resources, are the other migrated to other nodes?
> >
> > Not yet.
> > Such a feature is planned though.
>
On Thu, 2009-10-22 at 12:37 +0200, Andrew Beekhof wrote:
> > That's easy enough to understand ... and I can't do any fine-tuning,
> > i.e. suppose that 4 nodes of my 10 node cluster fail, and then come up
> > again. If all resources have equal score on all nodes (without
> > counting stickiness), t
On Wed, 2009-10-21 at 01:16 +0200, Dejan Muhamedagic wrote:
> > In that case, we would need to send two synchronised stonith actions to
> > two different stonith devices to succesfully stonith a node. Would this
> > be best solved at pacemaker or at RA level? (Assuming this isn't solved
> > at hard
Hi,
I was wondering, does the stonith support support servers with a
redundant powersupply on two different powergrids?
In that case, we would need to send two synchronised stonith actions to
two different stonith devices to succesfully stonith a node. Would this
be best solved at pacemaker or a
On Fri, 2009-10-16 at 12:32 +0200, Johan Verrept wrote:
> Other than that, I have no more ideas
perhaps ptest -L -s -V helps...
___
Pacemaker mailing list
Pacemaker@oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker
n Grüßen / with kind regards
>
> Torsten Schmidt
>
> -Ursprüngliche Nachricht-
> Von: Johan Verrept [mailto:johan.verr...@able.be]
> Gesendet: Freitag, 16. Oktober 2009 12:03
> An: pacemaker@oss.clusterlabs.org
> Betreff: Re: [Pacemaker] Problems with colocation/order w
I am unsure this is the source of your problems, but I think you have a
colocation restraint too much.
> colocation co.ms.drbd.mysqlstack_on_res.ip.mysql inf:
ms.drbd.mysqlstack res.ip.mysql
Why this?
You already force both ms.drbd.mysqlstack and res.ip.mysql onto
ms.drbd.mysqldb:Master, so the
On Thu, 2009-10-15 at 17:22 +0200, Dejan Muhamedagic wrote:
> Good patch, though it won't help you unless you create the
> directory which may not be exactly obvious, i.e. you would run
> into exactly the same problem,
I missed that. In my case the ${prefix}/var/lock dir exists already on
my syst
On Thu, 2009-10-15 at 17:16 +0200, Dejan Muhamedagic wrote:
> configure.ac says:
>
> for HA_VARLOCKDIR in ${localstatedir}/lock ${localstatedir}/spool/lock
> ${localstatedir}/spool/locks ${localstatedir}/lock
> do
> if
> test -d "$HA_VARLOCKDIR"
> then
> ...
>
> probably not s
On Thu, 2009-10-15 at 16:23 +0200, Johan Verrept wrote:
> For some reason, when building glue, HA_VARLOCKDIR is incorrectly
> #defined as "${prefix}/var/lock"
>
> ./include/glue_config.h:#define HA_VARLOCKDIR "${prefix}/var/lock"
The attached patch solves the p
On Thu, 2009-10-15 at 16:03 +0200, Johan Verrept wrote:
> On Thu, 2009-10-15 at 15:51 +0200, Johan Verrept wrote:
>
> > I just fear that some critical paths that are needed have moved between
> > 2.99.2 and 3.0.0 and that I am missing them. No idea how to find them,
> > ei
On Thu, 2009-10-15 at 15:51 +0200, Johan Verrept wrote:
> I just fear that some critical paths that are needed have moved between
> 2.99.2 and 3.0.0 and that I am missing them. No idea how to find them,
> either.
strace to the rescue...
4345 open("${prefix}/var/lock/LCK..ttyS0&qu
On Thu, 2009-10-15 at 15:49 +0200, Raoul Bhatia [IPAX] wrote:
> is another program using ttyS0? (e.g. getty? ser2net? etc.)
> (lsof -n |grep /dev/ttyS0)
nope.
> is it possible to *directly* communicate via the serial line?
It worked fine under heartbeat 2.99.2
> what happens if you connect the
On Thu, 2009-10-15 at 15:03 +0200, Dejan Muhamedagic wrote:
> Hi,
>
> On Thu, Oct 15, 2009 at 02:48:45PM +0200, Johan Verrept wrote:
> >
> > Hello,
> >
> > I don't understand why heartbeat/pacemaker does not want to start.
> > This is all logs ther
On Thu, 2009-10-15 at 15:06 +0200, Raoul Bhatia [IPAX] wrote:
> On 10/15/2009 02:48 PM, Johan Verrept wrote:
> >
> > Hello,
> >
> > I don't understand why heartbeat/pacemaker does not want to start.
> > This is all logs there are...
> >
> >
Hello,
I don't understand why heartbeat/pacemaker does not want to start.
This is all logs there are...
14:37:46 heartbeat [3010]: info: Version 2 support: yes
14:37:46 heartbeat [3010]: WARN: File /ub/etc/ha.d//haresources exists.
14:37:46 heartbeat [3010]: WARN: This file is not used because
On Thu, 2009-10-15 at 13:22 +0200, Andrew Beekhof wrote:
> > I found the --with-ocf-root parameter to configure in glue and
> > resource-agents so I can now build succesfully. This should still
> > default to the $PREFIX/usr/lib/ocf dir?
>
> More likely, $PREFIX/lib/ocf (assuming you insist on no
On Thu, 2009-10-15 at 10:12 +0200, Johan Verrept wrote:
> It installs the resources in /usr, not in $PREFIX/usr
> Is this intentional?
> It did not happen on hearbeat 2.99.2
I found the --with-ocf-root parameter to configure in glue and
resource-agents so I can now build succesfu
To see how node scores are calculated, try
ptest -L -s
J.
On Wed, 2009-10-14 at 19:07 +0300, Mihai wrote:
> Hello,
> I have a config like this:
> Online: [ int3.test.tst wor1.test.tst wor2.test.tst wor3.test.tst
> int1.test.tst int2.test.tst ]
>
> ip1 (ocf::heartbeat:IPaddr):
Hi Andrew,
On Wed, 2009-10-14 at 16:46 +0200, Andrew Beekhof wrote:
> > The distro is non-standard. I don't see how this would be different on
> > OpenSuSE? Do they use untagged (or specifically tagged) versions?
>
> I just built the lot (same versions as above) on OSX, and its fine here.
Ok...
gt; promoted to one of the nodes.
>
> Is this wrong? Should I try a different approach?
>
> Stratos.
>
> On Wed, Oct 14, 2009 at 1:02 PM, Johan Verrept
> wrote:
> I don't know if it is your only problem but in the order you
> refer to
>
I don't know if it is your only problem but in the order you refer to
primitive "mysql" while in your configuration you have primitive
"mysqld". Note the "d".
Other than that, the order constraint looks like mine.
J.
On Wed, 2009-10-14 at 12:59 +0300, Stratos Zolotas wrote:
> Hello again.
On Wed, 2009-10-14 at 10:58 +0200, Andrew Beekhof wrote:
> On Wed, Oct 14, 2009 at 9:25 AM, Johan Verrept wrote:
> > On Wed, 2009-10-14 at 08:21 +0200, Andrew Beekhof wrote:
> >> On Tue, Oct 13, 2009 at 6:04 PM, Johan Verrept
> >> wrote:
> >> >
> >&g
On Wed, 2009-10-14 at 08:21 +0200, Andrew Beekhof wrote:
> On Tue, Oct 13, 2009 at 6:04 PM, Johan Verrept wrote:
> >
> > Hi,
> >
> > when I install the latest resource-agents, it installs
> > etc/ha.d/shellfuncs
> >
> > This file includes @OCF
On Tue, 2009-10-13 at 18:04 +0200, Johan Verrept wrote:
> Hi,
>
> when I install the latest resource-agents, it installs
> etc/ha.d/shellfuncs
>
> This file includes @OCF_RA_DIR@/heartbeat/.ocf-shellfuncs at the end.
>
> .ocf-shellfuncs was in heartbeat 2.99.2, but
Hi,
when I install the latest resource-agents, it installs
etc/ha.d/shellfuncs
This file includes @OCF_RA_DIR@/heartbeat/.ocf-shellfuncs at the end.
.ocf-shellfuncs was in heartbeat 2.99.2, but is no longer present in
heartbeat 3.0.0. As far as I can tell this is now
lib/heartbeat/ocf-shellfunc
On Tue, 2009-10-13 at 17:49 +0200, Fausto Lombardi wrote:
> Hi of all,
> can I do a group where the resources in it start in sequence and if
> one of this resource fails, fail also the other in sequence?
This is how a "group" behaves.
See manual under "group"
J.
On Tue, 2009-10-13 at 13:52 +0200, Dejan Muhamedagic wrote:
> crm knows when the user's not in the interactive mood, so it may
> behave accordingly. Though the error message is still going to
> remain, it will be less obtrusive and go to stderr.
Awesome!
Thanks,
J.
___
Hi Dejan,
On Wed, 2009-10-07 at 17:06 +0200, Dejan Muhamedagic wrote:
> Yes, that's no problem, it's just that I'm not sure about how to
> design it since the language is, well, rather flat.
Might it be possible to at least let crm recognise the configuration as
valid or ignore it even if it isn'
On Mon, 2009-10-12 at 09:06 +0200, Andrew Beekhof wrote:
> On Mon, Oct 12, 2009 at 8:43 AM, Florian Haas wrote:
> > Andrew, Dejan: as we consider the ocf:linbit:drbd RA stable as of the
> > DRBD 8.3.4 release, is it acceptable to remove the legacy RA from the
> > agents repository?
>
> I think ra
On Tue, 2009-10-13 at 09:51 +0200, Florian Haas wrote:
> On 2009-10-13 09:40, Johan Verrept wrote:
> > On Mon, 2009-10-12 at 09:06 +0200, Andrew Beekhof wrote:
> >> On Mon, Oct 12, 2009 at 8:43 AM, Florian Haas
> >> wrote:
> >>> Andrew, Dejan: as we consi
On Sun, 2009-10-11 at 21:57 +0200, Andrew Beekhof wrote:
> just fixed it now, thanks
> also needed to change days -> weekdays and add a score to the rule itself.
It works now. Thank you for your help and confirmation.
J.
___
Pacemaker mailing
On Sun, 2009-10-11 at 21:33 +0200, Andrew Beekhof wrote:
> Then you didn't update glue properly :-)
> glue_config.h doesn't contain HA_LIBDIR anymore.
Correct. Updated glue and it builds fine now :)
On Mon, 2009-10-12 at 10:59 +0200, Dejan Muhamedagic wrote:
> This one's mine. Fixed today.
Confi
On Tue, 2009-10-13 at 09:52 +0200, Andrew Beekhof wrote:
> On Tue, Oct 13, 2009 at 9:40 AM, Johan Verrept wrote:
> > On Mon, 2009-10-12 at 09:06 +0200, Andrew Beekhof wrote:
> >> On Mon, Oct 12, 2009 at 8:43 AM, Florian Haas
> >> wrote:
> >> > Andrew, De
On Tue, 2009-10-06 at 17:15 +0200, Andrew Beekhof wrote:
> Anyway, I committed some changes to glue and hb3 which should fix the problem.
Apparantly, it didn't quite work. I upgraded to the current tips of
clusterglue, resource-agents and heartbeat. Pacemaker is still 1.0.5
pacemaker: In file i
On Thu, 2009-10-08 at 12:03 +0200, Johan Verrept wrote:
> thank you for answering. cibadmin does not want to accept the snippet
> though:
>
> Call cib_modify failed (-47): Update does not conform to the configured
> schema/DTD
I have been playing with this and it only applies
:
On Fri, Sep 25, 2009 at 11:21 AM, Johan Verrept wrote:
> >
> > Hi,
> >
> > I have seen this mentioned in the "Configuration Explained" manual and
> > it listed the ru
On Tue, 2009-10-06 at 16:28 +0200, Andrew Beekhof wrote:
> Its the same as HEAD from CVS.
Ah. So I am running ahead of releases :D
Good to know.
>
> What are the values for HA_LIBDIR and HA_HBCONF_DIR in
> /ub/include/heartbeat/hb_config.h and
/* lib directory */
#define HA_LIBDIR "/ub/lib"
/
On Tue, 2009-10-06 at 12:50 +0200, Dejan Muhamedagic wrote:
> Well, if we already have the option to use port-nodename mapping,
> that should be preferable to increasing the configuration size.
> Still, perhaps you could add a hostlist parameter like in most
> stonith plugins. Then just use that i
On Tue, 2009-10-06 at 15:57 +0200, Andrew Beekhof wrote:
> > I build straight from source. I downloaded the packages on
> > http://www.clusterlabs.org/wiki/Install#From_Source and more or less
> > followed instructions.
> >
> > The system I am building on isn't a standard distribution. It can build
Hi Lars,
> Yes.
> hostname (-s, -f) and uname differ in that
> uname -n just does one syscall (uname),
> and hostname (typically) does a gethostname(),
> then a gethostbyname(), the result of the latter
> heavily influenced by whatever is in your nsswitch,
> resolv and other .conf, as well as wha
On Mon, 2009-10-05 at 16:30 +0200, Colin wrote:
> - The pacemaker-heartbeat installation, with "crm yes" in
> /etc/ha.d/ha.cf, does not seem to start pacemaker, at least a crm_mon
> just hangs trying to connect to the cluster (the documentation doesn't
> give any hint on how to check), and a littl
I don't think it is supposed to work that way. Nodes are cluster nodes.
If you want to ping a machine and use that in your cluster, see
http://clusterlabs.org/wiki/Example_configurations#Set_up_pingd
J.
On Mon, 2009-10-05 at 16:14 +0200, Fausto Lombardi wrote:
> Hi of all,
> I'm trying to
Hi,
when playing with the RA at a certain point the stonith failed (it
didn't find the host in gethosts) and I rebooted the other node
manually. The result was a whole bunch of messages in my logs:
15:53:10 SYSLOG warning heartbeat [2748]: WARN: Gmain_timeout_dispatch:
Dispatch function for
On Mon, 2009-10-05 at 14:38 +0200, Johan Verrept wrote:
> Perhaps something in the nature of
>
> if [ `uname -n` = `hostname -s` ]
> then
> for host in `list_devices`;
> do
> echo $host
> done;
> else
> domainname=`ho
On Mon, 2009-10-05 at 14:01 +0200, Florian Haas wrote:
> > gethosts)
> > # print hosts with domainname
> > domain=`domainname`
> > for host in `list_devices`;
> > do
> > echo $host.$domain
> > done;
>
> Can you make sure that this works both on systems where "uname
allow it to reset ports. You can use the default
admin
# account or create an account for the RA through the web interface
of
# the device.
#
On Mon, 2009-10-05 at 12:42 +0200, Dejan Muhamedagic wrote:
> Hi,
>
> On Mon, Oct 05, 2009 at 10:37:54AM +0200, Johan Verrept wrote:
Hello guys,
I completed the RA and have attached it. As far as I can tell it is
fully functional but I would appreciate it if someone "in the know"
would take a look and gave me some feedback.
Thank you,
Johan Verrept
On Fri, 2009-10-02 at 11:04 +0200, Johan Verrept w
On Fri, 2009-10-02 at 16:05 +0200, Dejan Muhamedagic wrote:
> >
> > the resource-agents package file heartbeat shellfuncs contains
> > @HA_NOARCHDATAHBDIR@ but this is not replaced. (and not defined in the
> > makefiles).
>
> Indeed, @HA_NOARCHDATAHBDIR@ is not replaced, but that shouldn't
>
On Fri, 2009-10-02 at 16:02 +0200, Dejan Muhamedagic wrote:
> Hi,
>
> On Fri, Oct 02, 2009 at 01:55:47PM +0200, Johan Verrept wrote:
> > Hello,
> >
> >I am trying to build pacemaker 1.0.5, I get the following error:
>
> How do you try to build? Using r
Dejan,
thanks for your answer. I will repeat the test on monday and provide
logs.
Regards,
Johan
On Fri, 2009-10-02 at 15:52 +0200, Dejan Muhamedagic wrote:
> Hi,
>
> On Fri, Oct 02, 2009 at 11:18:49AM +0200, Johan Verrept wrote:
> >
> > Hi,
> >
&
Hi,
the resource-agents package file heartbeat shellfuncs contains
@HA_NOARCHDATAHBDIR@ but this is not replaced. (and not defined in the
makefiles).
clusterglue 1.0.0
resource-agents 1.0.0
heartbeat 3.0.0
pacemaker 1.0.5
this is the second build error.
Am I usin
Hello,
I am trying to build pacemaker 1.0.5, I get the following error:
make[2]: Entering directory
`/root/work/Pacemaker-1-0-Pacemaker-1.0.5/fencing/stonithd'
if gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I. -I../../include
-I../../include -I../../include -I../../include -I../../libltdl
-I../
Hi,
I have a master/slave cluster configured with stonith. When I
ifconfig down all interfaces on the slave node, the other node stoniths
it. This succeeds and the node comes back to the cluster succesfully.
Then, however, it seems the cluster stoniths the master node... All I
know for sure is
+0200, Johan Verrept wrote:
> Hi guys,
>
> When I looked into this yesterday , I found this device:
>
> http://www.aviosys.com/ip_power_9258hp.html
>
> At around 150 euro, it is pretty cheap. It is an independent device and
> will not do nifty things like server
On Sun, 2009-09-27 at 16:32 +1000, Matthew Palmer wrote:
> On a related topic, is there any way to find out what the cluster's scores
> for all resources are, and how it came to calculate those scores? The logs
> are full of trivia, but they lack this sort of really useful detail. I
> assume ther
Hi,
I have seen this mentioned in the "Configuration Explained" manual and
it listed the rules to use, but it didn't specify how to actually apply
that rule to the stickiness attribute. I have looked with google and
through the crm help but that didn't do me much good either.
Can I create thi
Hi guys,
When I looked into this yesterday , I found this device:
http://www.aviosys.com/ip_power_9258hp.html
At around 150 euro, it is pretty cheap. It is an independent device and
will not do nifty things like server console access (like the IPMI
cards) but it should work well for ston
If I use the link you poster earlier in this thread, things look normal
here with xpdf, 1600x1200 and using "fit width".
I had the same problem with the manual that has those cluster
transitions diagrams in it. They are barely readable and even when
zooming they are blurry and hard to read.
Hi Lars,
yes, thank you, much more comfortable :)
Regards,
Johan
PS: Sorry for the late reply, stubborn flu thingie.
> Would that be better?
>
> http://git.drbd.org/?p=drbd-8.3.git;a=commitdiff;h=e937117d33edd1c510fb5dd381255d57bda6afea
>
>
__
On Tue, 2009-09-08 at 17:01 +0200, Florian Haas wrote:
> > As mentioned in my other post, it works as long as you do not have
> > multiple reasons to move your master. If you do, you limit the priority
> > space. This is not a problem in simple cases but it will be in more
> > complex configuration
On Tue, 2009-09-08 at 15:37 +0200, Florian Haas wrote:
> Because crm_master is already there, it works, it is what is designed to
> be used as an interface for changing master preferences
As mentioned in my other post, it works as long as you do not have
multiple reasons to move your master. If y
On Tue, 2009-09-08 at 15:04 +0200, Florian Haas wrote:
> The question is moot in this case, though, as I already mentioned: all
> of this is handled via master preferences which the ocf:linbit:drbd RA
> updates automatically.
Yes. This works nicely with the new version. One remark I have is that
Hello,
I have been thinking about how the DRBD RA influences the master
placement by manipulating a temporary attribute. How does that work
exactly? Does the crm interpret certain attributes? Does it work on all
numerical attributes? If so, does this not limit their use?
Why isn't a location rule
On Fri, 2009-09-04 at 14:32 +0200, Florian Haas wrote:
> If it's not of any concern to you that your hard failover is not
> automatic, why not go with the meatware "plugin" (aka human
> intervention) in the first place?
Unless I misunderstood his setup, only when the power to the other half
the da
Hello Florian,
thank you for your help, the ptest commands was very useful. Perhaps
this could be mentioned in the Pacemaker wiki under a Troubleshooting
section? I read all the docs and didn't see it mentioned anywhere (could
have missed it).
I see now what happens without the rule. I have an in
Hello,
I solved my problems with the DRBD RA and now the attributes are set
correctly in the CIB and my location rules seem to apply. Now the
problem seems to be that the location rule doesn't do what I expect it
to.
crm configure location ms-drbd0-primary-never-on-synctarget ms-drbd0 \
rule
Hi Florian,
thank you for answering so swiftly!
> >I am using the heartbeat::drbd OCF RA.
>
> That's the problem. Use DRBD 8.3.2, and use the ocf:linbit:drbd resource
> agent that's bundled with it.
I upgraded drbd and switched to that RA. As far as I can see, the values
assigned to the n
Hello,
I have been trying to keep my cluster from making the synctarget the
drbd primary, but I cannot get it to work. See the end of the email for
versions I use.
I am using the heartbeat::drbd OCF RA. I had to patch it so it would
work with my drbd version (see below). As far as I can tell
74 matches
Mail list logo