[Linux-cluster] RHCS + DRBD - DRBD file systems not mounting

[Goutam Baul]

Dear List,

 

We have configured DRBD 8.4.3 (community version) in the sense that we are
able to manually  start DRBD at both the nodes, make nodes primary or
secondary and data is getting replicated between the primary and the
secondary node. We are doing this for our postfix mail server. The drbd
configuration file is

 

resource r0 {

startup {

    wfc-timeout 1;

    degr-wfc-timeout 2;

  }

  volume 0 {

    device    /dev/drbd1;

    disk      /dev/emcpowera2;

    meta-disk /dev/emcpowera5;

  }

  volume 1 {

    device    /dev/drbd2;

    disk      /dev/emcpowera3;

    meta-disk /dev/emcpowera6;

  }

  on drmail1.cesc.co.in {

    address   10.50.4.14:7789;

  }

  on drbdtest2.cesc.co.in {

    address   10.50.81.253:7789;

  }

}

 

We want to run postfix mailing system on a RHCS cluster suite for high
availability. The mail-store and the mail-queue of postfix will be on the
DRBD setup so that they are replicated from our data center to our disaster
recovery site. The cluster.conf is as follows:

 

<?xml version="1.0"?>

<cluster config_version="27" name="dr-mail">

        <fence_daemon clean_start="1" post_join_delay="30"/>

        <clusternodes>

                <clusternode name="drmail1.cesc.co.in" nodeid="1">

                        <fence>

                                <method name="Method1_drac">

                                        <device name="fence_drac1"/>

                                </method>

 

                        </fence>

                </clusternode>

                <clusternode name="drmail2.cesc.co.in" nodeid="2">

                        <fence>

                                <method name="Method1_drac">

                                        <device name="fence_drac2"/>

                                </method>

 

                        </fence>

                </clusternode>

        </clusternodes>

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

        <fencedevices>

                <fencedevice agent="fence_drac5" cmd_prompt="admin1->"
ipaddr="10.50.4.22" login="root" module_name="fence_drac1"
name="fence_drac1" passwd="calvin" secure="on"/>

                <fencedevice agent="fence_drac5" cmd_prompt="admin1->"
ipaddr="10.50.4.23" login="root" module_name="fence_drac2"
name="fence_drac2" passwd="calvin" secure="on"/>

 

        </fencedevices>

        <rm>

                <failoverdomains>

                        <failoverdomain name="dr_mail" nofailback="0"
ordered="1" restricted="1">

                                <failoverdomainnode
name="drmail1.cesc.co.in" priority="1"/>

                                <failoverdomainnode
name="drmail2.cesc.co.in" priority="2"/>

                        </failoverdomain>

                </failoverdomains>

                <resources>

                        <ip address="10.50.4.20" monitor_link="on"
sleeptime="10"/>

                        <drbd name="drbd-mail" resource="r0">

                              <fs device="/dev/drbd/by-res/r0/0"
force_unmount="1" fsid="61850" mountpoint="/home" fstype="ext3"
name="drbd-mailstore" options="noatime"/>

                              <fs device="/dev/drbd/by-res/r0/1"
force_unmount="1" fsid="61855" mountpoint="/var/spool/postfix" fstype="ext3"
name="drbd-spool" options="noatime"/>

                        </drbd>

                        <script file="/etc/init.d/postfix" name="postfix" />

                </resources>

                <service domain="dr_mail" name="mail" recovery="relocate">

                        <ip ref="10.50.4.20"/>

                        <drbd ref="drbd-mail" />

                        <script ref="postfix" />

                </service>

        </rm>

</cluster>

 

If we run this RHCS cluster, we get the error in the log as 

 

Aug 27 17:38:19 drmail1 rgmanager[4695]: drbd not found in /proc/modules. Do
you need to modprobe?

 

If we run the RHCS cluster after issuing the modprobe drbd command we get
the following message

 

1: Failure: (127) Device minor not allocated

additional info from kernel:

unknown minor

Command 'drbdsetup role 1' terminated with exit code 10

<debug>  DRBD resource r0 already configured

DRBD resource r0 already configured

Start of mail complete

 

But the DRBD service is not starting up and therefore the corresponding
devices are not being created. Thus the log file returns the error

 

Aug 27 17:38:33 drmail1 rgmanager[5147]: stop: Could not match
/dev/drbd/by-res/r0/1 with a real device

Aug 27 17:38:33 drmail1 rgmanager[5166]: stop: Could not match
/dev/drbd/by-res/r0/0 with a real device

 

We have built the DRBD rpm files from the source using rpmbuild and copied
the drbd.sh.rhcs and drbd.metadata.rhcs files from the script folder of the
source to the /usr/share/cluster folder as drbd.sh and drbd.metadata
respectively. We are using RHEL 6.1 (64 bit).

 

Any help please.

 

With regards,

 

Goutam

 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20130827/1000014a/attachment.htm>