Mailing List Archive: DRBD: Users

drbd with gfs2

Index | Next | Previous | View Threaded

hijinks at gmail

May 13, 2008, 9:57 AM

Post #1 of 8 (3385 views)
Permalink

drbd with gfs2

We are having some issues with gfs2 and drbd. When both nodes are up things
work great. I can access the file on both nodes and they are in
primary/primary mode.

The issue is that if a node is down I cannot access the mount on the system
that is up. I can only access it when the node comes back online.

I followed a howto at the following URL just used a bit different drbd.conf

http://gfs.wikidev.net/DRBD_Cookbook

Below is my drbd.conf

global {
usage-count no;
}

common {
syncer { rate 100M; }
}

resource "drbd0" {
protocol C;

handlers {
pri-on-incon-degr "echo 'DRBD: primary requested but inconsistent!' | wall;
/etc/init.d/heartbeat stop"; #"halt -f";
pri-lost-after-sb "echo 'DRBD: primary requested but lost!' | wall;
/etc/init.d/heartbeat stop"; #"halt -f";
}

startup {
wfc-timeout 0; # 2 minutes
degr-wfc-timeout 120; # 2 minutes.
become-primary-on both;
}

syncer {
rate 100M;
# This is now expressed with "after res-name"
#group 1;
al-extents 257;
}

net {
allow-two-primaries;
after-sb-0pri discard-zero-changes;
after-sb-1pri discard-secondary;
after-sb-2pri disconnect;

}

disk {
on-io-error pass_on;
fencing dont-care;
}

on njrh-testdfs1.fxserver.com {
device /dev/drbd0;
disk /dev/hdb1;
address 10.100.172.195:7788;
meta-disk internal;
}

on njrh-testdfs2.fxserver.com {
device /dev/drbd0;
disk /dev/hdb1;
address 10.100.172.196:7788;
meta-disk internal;
}
}

drbd at bobich

May 13, 2008, 10:07 AM

Post #2 of 8 (3351 views)
Permalink

Re: drbd with gfs2 [In reply to]

On Tue, 13 May 2008, Mike Zupan wrote:

> We are having some issues with gfs2 and drbd. When both nodes are up
> things work great. I can access the file on both nodes and they are in
> primary/primary mode.
>
> The issue is that if a node is down I cannot access the mount on the
> system that is up. I can only access it when the node comes back online.

>From your description, this is not a DRBD problem. It sounds like you
haven't set your fencing up correctly. GFS/GFS2 blocks all FS access until
the gencing script reports that the node has been powered down and
verified as such.

Gordan
_______________________________________________
drbd-user mailing list
drbd-user [at] lists
http://lists.linbit.com/mailman/listinfo/drbd-user

hijinks at gmail

May 13, 2008, 10:13 AM

Post #3 of 8 (3344 views)
Permalink

Re: drbd with gfs2 [In reply to]

Any examples of what should be setup for fencing.. I just used the example
in the howto

Mike

On 5/13/08, drbd [at] bobich <drbd [at] bobich> wrote:
>
>
>
> On Tue, 13 May 2008, Mike Zupan wrote:
>
> We are having some issues with gfs2 and drbd. When both nodes are up
> > things work great. I can access the file on both nodes and they are in
> > primary/primary mode.
> >
> > The issue is that if a node is down I cannot access the mount on the
> > system that is up. I can only access it when the node comes back online.
> >
>
> From your description, this is not a DRBD problem. It sounds like you
> >
> haven't set your fencing up correctly. GFS/GFS2 blocks all FS access until
> the gencing script reports that the node has been powered down and verified
> as such.
>
> Gordan
> _______________________________________________
> drbd-user mailing list
> drbd-user [at] lists
> http://lists.linbit.com/mailman/listinfo/drbd-user
>

drbd at bobich

May 13, 2008, 10:21 AM

Post #4 of 8 (3334 views)
Permalink

Re: drbd with gfs2 [In reply to]

You need to set up your fencing devices in /etc/cluster/cluster.conf.

In each <clusternode> section you need a block like
<fence>
<method name = "foo">
<device name = "fence_foo"/>
</method>
</fence>

And then after </clusternodes> a
block like:

<fencedevices>
<fencedevice ........>
<fencedevice ........>
</fencedices>

where dots will be dependant on what fencing devices you are using (IPMI,
Dell DRAC, HP iLO, APC power bars/UPS, etc.)

Gordan

On Tue, 13 May 2008, Mike Zupan wrote:

> Any examples of what should be setup for fencing.. I just used the example in the howto
>
> Mike
>
> On 5/13/08, drbd [at] bobich <drbd [at] bobich> wrote:
>
>
> On Tue, 13 May 2008, Mike Zupan wrote:
>
> We are having some issues with gfs2 and drbd. When both nodes are up
> things work great. I can access the file on both nodes and they are
> in primary/primary mode.
>
> The issue is that if a node is down I cannot access the mount on the
> system that is up. I can only access it when the node comes back
> online.
>
>
> From your description, this is not a DRBD problem. It sounds like you
>
> haven't set your fencing up correctly. GFS/GFS2 blocks all FS access until the
> gencing script reports that the node has been powered down and verified as such.
>
> Gordan
> _______________________________________________
> drbd-user mailing list
> drbd-user [at] lists
> http://lists.linbit.com/mailman/listinfo/drbd-user
>
>
>
>
_______________________________________________
drbd-user mailing list
drbd-user [at] lists
http://lists.linbit.com/mailman/listinfo/drbd-user

hijinks at gmail

May 13, 2008, 10:40 AM

Post #5 of 8 (3345 views)
Permalink

Re: drbd with gfs2 [In reply to]

On 5/13/08, drbd [at] bobich <drbd [at] bobich> wrote:
>
> You need to set up your fencing devices in /etc/cluster/cluster.conf.
>
> In each <clusternode> section you need a block like
> <fence>
> <method name = "foo">
> <device name = "fence_foo"/>
> </method>
> </fence>
>
> And then after </clusternodes> a block like:
>
> <fencedevices>
> <fencedevice ........>
> <fencedevice ........>
> </fencedices>

It seems I do have all that setup

<?xml version="1.0"?>

<cluster alias="lolcats" config_version="41" name="lolcats">

<cman expected_votes="2" two_node="1"/>

<clusternodes>

<clusternode name="njrh-testdfs1.fxserver.com" nodeid="1" votes="1">

<fence>

<method name="1">

<device domain="njrh-testdfs1.fxserver.com" name="xvm"/>

</method>

</fence>

</clusternode>

<clusternode name="njrh-testdfs2.fxserver.com" nodeid="2" votes="1">

<fence>

<method name="1">

<device domain="njrh-testdfs2.fxserver.com" name="xvm"/>

</method>

</fence>

</clusternode>

</clusternodes>

<fencedevices>

<fencedevice agent="fence_xvm" name="xvm"/>

</fencedevices>

<rm/>

</cluster>

drbd at bobich

May 13, 2008, 11:11 AM

Post #6 of 8 (3338 views)
Permalink

Re: drbd with gfs2 [In reply to]

Check what /var/log/messages on the remaining node says. It will tell
you if/why fencing is failing.

Gordan

Mike Zupan wrote:
>
>
> On 5/13/08, *drbd [at] bobich <mailto:drbd [at] bobich>* <drbd [at] bobich
> <mailto:drbd [at] bobich>> wrote:
>
> You need to set up your fencing devices in /etc/cluster/cluster.conf.
>
> In each <clusternode> section you need a block like
> <fence>
> <method name = "foo">
> <device name = "fence_foo"/>
> </method>
> </fence>
>
> And then after </clusternodes> a block like:
>
> <fencedevices>
> <fencedevice ........>
> <fencedevice ........>
> </fencedices>
>
>
> It seems I do have all that setup
>
>
> <?xml version="1.0"?>
>
> <cluster alias="lolcats" config_version="41" name="lolcats">
>
> <cman expected_votes="2" two_node="1"/>
>
> <clusternodes>
>
> <clusternode name="njrh-testdfs1.fxserver.com
> <http://njrh-testdfs1.fxserver.com>" nodeid="1" votes="1">
>
> <fence>
>
> <method name="1">
>
> <device domain="njrh-testdfs1.fxserver.com
> <http://njrh-testdfs1.fxserver.com>" name="xvm"/>
>
> </method>
>
> </fence>
>
> </clusternode>
>
> <clusternode name="njrh-testdfs2.fxserver.com
> <http://njrh-testdfs2.fxserver.com>" nodeid="2" votes="1">
>
> <fence>
>
> <method name="1">
>
> <device domain="njrh-testdfs2.fxserver.com
> <http://njrh-testdfs2.fxserver.com>" name="xvm"/>
>
> </method>
>
> </fence>
>
> </clusternode>
>
> </clusternodes>
>
> <fencedevices>
>
> <fencedevice agent="fence_xvm" name="xvm"/>
>
> </fencedevices>
>
> <rm/>
>
> </cluster>
>
>
>
>

_______________________________________________
drbd-user mailing list
drbd-user [at] lists
http://lists.linbit.com/mailman/listinfo/drbd-user

hijinks at gmail

May 13, 2008, 1:06 PM

Post #7 of 8 (3330 views)
Permalink

Re: drbd with gfs2 [In reply to]

It really doesn't say anything.. just fencing has failed to the other down
node

On 5/13/08, Gordan Bobic <drbd [at] bobich> wrote:
>
> Check what /var/log/messages on the remaining node says. It will tell you
> if/why fencing is failing.
>
> Gordan
>
> Mike Zupan wrote:
>
> >
> >
> > On 5/13/08, *drbd [at] bobich <mailto:drbd [at] bobich>* <drbd [at] bobich<mailto:
> > drbd [at] bobich>> wrote:
> >
> > You need to set up your fencing devices in /etc/cluster/cluster.conf.
> >
> > In each <clusternode> section you need a block like
> > <fence>
> > <method name = "foo">
> > <device name = "fence_foo"/>
> > </method>
> > </fence>
> >
> > And then after </clusternodes> a block like:
> >
> > <fencedevices>
> > <fencedevice ........>
> > <fencedevice ........>
> > </fencedices>
> >
> >
> > It seems I do have all that setup
> >
> >
> > <?xml version="1.0"?>
> >
> > <cluster alias="lolcats" config_version="41" name="lolcats">
> >
> > <cman expected_votes="2" two_node="1"/>
> >
> > <clusternodes>
> >
> > <clusternode name="njrh-testdfs1.fxserver.com <
> > http://njrh-testdfs1.fxserver.com>" nodeid="1" votes="1">
> >
> > <fence>
> >
> > <method name="1">
> >
> > <device domain="njrh-testdfs1.fxserver.com <
> > http://njrh-testdfs1.fxserver.com>" name="xvm"/>
> >
> > </method>
> >
> > </fence>
> >
> > </clusternode>
> >
> > <clusternode name="njrh-testdfs2.fxserver.com <
> > http://njrh-testdfs2.fxserver.com>" nodeid="2" votes="1">
> >
> > <fence>
> >
> > <method name="1">
> >
> > <device domain="njrh-testdfs2.fxserver.com <
> > http://njrh-testdfs2.fxserver.com>" name="xvm"/>
> >
> > </method>
> >
> > </fence>
> >
> > </clusternode>
> >
> > </clusternodes>
> >
> > <fencedevices>
> >
> > <fencedevice agent="fence_xvm" name="xvm"/>
> >
> > </fencedevices>
> >
> > <rm/>
> >
> > </cluster>
> >
> >
> >
> >
> _______________________________________________
> drbd-user mailing list
> drbd-user [at] lists
> http://lists.linbit.com/mailman/listinfo/drbd-user
>

drbd at bobich

May 13, 2008, 1:20 PM

Post #8 of 8 (3340 views)
Permalink

Re: drbd with gfs2 [In reply to]

Well, that's why GFS stops working. You need to work out why it's
failing. I've never used that particular fencing device, so I can't help
you, but checking through the documentation for the xvm fencing agent
might be a good start. If that doesn't tell you what you need, try
asking on the redhat clustering (RHCE) mailing list. That would be the
right place to ask, rather than here as it has nothing at all to do with
DRBD.

Gordan

Mike Zupan wrote:
> It really doesn't say anything.. just fencing has failed to the other
> down node
>
> On 5/13/08, *Gordan Bobic* <drbd [at] bobich <mailto:drbd [at] bobich>> wrote:
>
> Check what /var/log/messages on the remaining node says. It will
> tell you if/why fencing is failing.
>
> Gordan
>
> Mike Zupan wrote:
>
>
>
> On 5/13/08, *drbd [at] bobich <mailto:drbd [at] bobich>
> <mailto:drbd [at] bobich <mailto:drbd [at] bobich>>*
> <drbd [at] bobich <mailto:drbd [at] bobich>
> <mailto:drbd [at] bobich <mailto:drbd [at] bobich>>> wrote:
>
> You need to set up your fencing devices in
> /etc/cluster/cluster.conf.
>
> In each <clusternode> section you need a block like
> <fence>
> <method name = "foo">
> <device name = "fence_foo"/>
> </method>
> </fence>
>
> And then after </clusternodes> a block like:
>
> <fencedevices>
> <fencedevice ........>
> <fencedevice ........>
> </fencedices>
>
>
> It seems I do have all that setup
>
>
> <?xml version="1.0"?>
>
> <cluster alias="lolcats" config_version="41" name="lolcats">
>
> <cman expected_votes="2" two_node="1"/>
>
> <clusternodes>
>
> <clusternode name="njrh-testdfs1.fxserver.com
> <http://njrh-testdfs1.fxserver.com>
> <http://njrh-testdfs1.fxserver.com>" nodeid="1" votes="1">
>
> <fence>
>
> <method name="1">
>
> <device domain="njrh-testdfs1.fxserver.com
> <http://njrh-testdfs1.fxserver.com>
> <http://njrh-testdfs1.fxserver.com>" name="xvm"/>
>
> </method>
>
> </fence>
>
> </clusternode>
>
> <clusternode name="njrh-testdfs2.fxserver.com
> <http://njrh-testdfs2.fxserver.com>
> <http://njrh-testdfs2.fxserver.com>" nodeid="2" votes="1">
>
> <fence>
>
> <method name="1">
>
> <device domain="njrh-testdfs2.fxserver.com
> <http://njrh-testdfs2.fxserver.com>
> <http://njrh-testdfs2.fxserver.com>" name="xvm"/>
>
> </method>
>
> </fence>
>
> </clusternode>
>
> </clusternodes>
>
> <fencedevices>
>
> <fencedevice agent="fence_xvm" name="xvm"/>
>
> </fencedevices>
>
> <rm/>
>
> </cluster>
>
>
>
>
> _______________________________________________
> drbd-user mailing list
> drbd-user [at] lists <mailto:drbd-user [at] lists>
> http://lists.linbit.com/mailman/listinfo/drbd-user
>
>

_______________________________________________
drbd-user mailing list
drbd-user [at] lists
http://lists.linbit.com/mailman/listinfo/drbd-user

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads