Hello again,
I have saw your configuration and it is the same as I was using in the
drbd configuration part. I didn't have a group now, I will create one
for Filesystem, IPAddress and other services who depends of drbd
storage. My goal is improve the HA of a mail server because now the drbd
failover fails ramdonly (I am using drbddisk but I want to forget it and
use a Master/Slave better configuration).
One doubt about your config is the next: you have stonith enabled but
you haven't defined a stonith resource the you couldn't use fence. Have
you thinked about it ?
Using your config with the drbd official script in the slave node the
drbd isn't started but the monitor said that it is started. A
improvement is drbd script supports the two resources, with my config I
can't make it works with two resources. With my script all is running
OK, I don't know why but I am not implementing notify action in my
script, and official script seems to fail at this action.
Have anyone else tried official script with two resources ? Anyone else
who has problems with official script ? Another who has tried my script ?
Serge Dubrouski escribió:
On Wed, Oct 22, 2008 at 11:53 AM, Adrian Chapela
<[EMAIL PROTECTED]> wrote:
Thank you!
I will have a look tomorrow. For me is the end of the day today.
Have you used the official drbd with your config ? drbd OCF script is ok
for you ?
Yes I use the official one and it works all right for me.
Ok, Have you tried with two drbd resources ? I think this is the problem
of the drbd official script. Also, have you saw if the slave node is
started (have the drbd module loaded, etc...)
Thank you again.
Serge Dubrouski escribió:
I attached it to my previous e-mail, but here it is again:
<cib>
<configuration>
<crm_config>
<cluster_property_set id="cib-bootstrap-options">
<nvpair id="global-resource-stickiness"
name="default-resource-stickiness" value="600"/>
<nvpair id="global-resource-failure-stickiness"
name="default-resource-failure-stickiness" value="-520"/>
<nvpair id="symmetric-cluster" name="symmetric-cluster"
value="true"/>
<nvpair id="stonith-enabled" name="stonith-enabled" value="true"/>
<nvpair id="dc-version" name="dc-version" value="0.7.1-node:
81043207472250ea93afed81a05f35253ffb8122"/>
</cluster_property_set>
</crm_config>
<nodes>
<node id="b88f98c6-50f2-463a-a6eb-51abbec645a9" uname="fc-node1"
type="normal"/>
<node id="ad6f19b7-228a-48b7-bae0-f95a838bde2a" uname="fc-node2"
type="normal"/>
</nodes>
<resources>
<master id="ms-drbd0">
<meta_attributes id="drbd_meta_attributes">
<nvpair id="drbd_clone_max" name="clone_max" value="2"/>
<nvpair id="drbd_clone_node_max" name="clone_node_max"
value="1"/>
<nvpair id="drbd_master_max" name="master_max" value="1"/>
<nvpair id="drbd_master_node_max" name="master_node_max"
value="1"/>
<nvpair id="drbd_notify" name="notify" value="yes"/>
<nvpair id="drbd_globally-unique" name="globally_unique"
value="false"/>
</meta_attributes>
<primitive id="drbd0" class="ocf" provider="heartbeat" type="drbd">
<instance_attributes id="drbd0_instance_attributes">
<nvpair id="drbd0_drbd_resource" name="drbd_resource"
value="drbd0"/>
</instance_attributes>
<operations>
<op name="monitor" interval="59s" id="op-drbd0-1"
timeout="10s" role="Master"/>
<op name="monitor" interval="60s" id="op-drbd0-2"
timeout="10s" role="Slave"/>
</operations>
</primitive>
</master>
<group id="myGroup">
<primitive id="myIP" class="ocf" type="IPaddr"
provider="heartbeat">
<instance_attributes id="myIP_instance_attributes">
<nvpair id="myIP_ip" name="ip" value="192.168.1.130"/>
</instance_attributes>
<operations>
<op name="monitor" interval="30s" id="myIP_monitor"
timeout="30s"/>
<op name="start" interval="0s" id="myIP_start" timeout="30s"/>
<op name="stop" interval="0s" id="myIP_stop" timeout="30s"/>
</operations>
</primitive>
<primitive id="fs0" class="ocf" provider="heartbeat"
type="Filesystem">
<instance_attributes id="fs0_instance_attributes">
<nvpair id="fs0_fstype" name="fstype" value="ext2"/>
<nvpair id="fs0_directory" name="directory" value="/mnt"/>
<nvpair id="fs0_device" name="device" value="/dev/drbd0"/>
</instance_attributes>
<operations>
<op name="monitor" interval="30s" id="fs0_monitor"
timeout="30s"/>
<op name="start" interval="0s" id="fs0_start" timeout="30s"/>
<op name="stop" interval="0s" id="fs0_stop" timeout="30s"/>
</operations>
</primitive>
<primitive id="myPgsql" class="ocf" type="pgsql"
provider="heartbeat">
<instance_attributes id="myPgsql_instance_attributes">
<nvpair id="myPgsql_ctl_opt" name="ctl_opt" value="-w"/>
</instance_attributes>
<operations>
<op name="monitor" interval="30s" id="pgsql_monitor"
timeout="30s"/>
<op name="start" interval="0s" id="pgsql_start" timeout="30s"/>
<op name="stop" interval="0s" id="pgsal_stop" timeout="30s"/>
</operations>
</primitive>
</group>
</resources>
<constraints>
<rsc_order id="drbd0_before_myGroup" first="ms-drbd0"
then="myGroup" then-action="start" first-action="promote"/>
<rsc_colocation id="myGroup_on_drbd0" rsc="myGroup"
with-rsc="ms-drbd0" with-rsc-role="Master" score="INFINITY"/>
<rsc_location id="primNode" rsc="myGroup">
<rule id="prefered_primNode" score="1000">
<expression attribute="#uname" id="expression.id2242728"
operation="eq" value="fc-node1"/>
</rule>
</rsc_location>
<rsc_location id="PTE-connected" rsc="myGroup">
<rule id="PGSQL-connected-rule" score="-INFINITY" boolean-op="or">
<expression id="expression.id2242755" attribute="pingd"
operation="not_defined"/>
<expression id="expression.id2242765" attribute="pingd"
operation="lte" value="0"/>
</rule>
</rsc_location>
</constraints>
</configuration>
</cib>
On Wed, Oct 22, 2008 at 8:32 AM, Adrian Chapela
<[EMAIL PROTECTED]> wrote:
Serge Dubrouski escribió:
Yes I can give it a try but the truth is that I never had problems
with the default one. I can give you my cib.xml if you want.
If you can I will be very happy it you send me your cib.xml
Thank you!
On Wed, Oct 22, 2008 at 5:02 AM, Adrian Chapela
<[EMAIL PROTECTED]> wrote:
Serge Dubrouski escribió:
On Tue, Oct 21, 2008 at 4:12 AM, Adrian Chapela
<[EMAIL PROTECTED]> wrote:
Again I am testing with a more simple config file, with only one
master/slave resource.
In this case the resource becomes to be a Master but in the slave
server
the
DRBD is not running. The modules is not loaded and crm_mon says that
drbd
is
running as Slave:
Master/Slave Set: ms-drbd0
drbd0:0 (ocf::heartbeat:drbd): Master debianquagga2
drbd0:1 (ocf::heartbeat:drbd): Started debianquagga1
Resource configuration:
<resources>
<master_slave id="ms-drbd0">
<meta_attributes id="ma-ms-drbd0">
<attributes>
<nvpair id="ma-ms-drbd0-1"
name="clone_max" value="2"/>
<nvpair id="ma-ms-drbd0-2"
name="clone_node_max" value="1"/>
<nvpair id="ma-ms-drbd0-3"
name="master_max" value="1"/>
<nvpair id="ma-ms-drbd0-4"
name="master_node_max" value="1"/>
<nvpair id="ma-ms-drbd0-5"
name="notify" value="yes"/>
<nvpair id="ma-ms-drbd0-6"
name="globally_unique" value="false"/>
</attributes>
</meta_attributes>
<primitive id="drbd0" class="ocf"
provider="heartbeat"
type="drbd">
<instance_attributes id="ia-drbd0">
<attributes>
<nvpair id="ia-drbd0-1"
name="drbd_resource" value="mail_disk"/>
</attributes>
</instance_attributes>
<operations>
<op id="op-ms-drbd2-1"
name="monitor"
interval="59s" timeout="60s" start_delay="30s" role="Master"/>
<op id="op-ms-drbd2-2"
name="monitor"
interval="60s" timeout="60s" start_delay="30s" role="Slave"/>
</operations>
</primitive>
</master_slave>
</resources>
Why heartbeat is not monitoring the service in the slave node ?
It does. I think that your slave is in Standalone mode in DRBD, check
it with drbdadm cstate.
Adrian Chapela escribió:
Hello,
I am doing new tests. I am doing this tests to improve an old config
and
to try to understand best multistate resources.
I can't make it to works, my configuration is always bad.... I
don't
know
anything. I am using 2.1.4 release.
The cluster is only doing a notify action and then stop action. If I
started drbd myself, the cluster bring drbd in an Uncofigured
resources
state. I don't know the reason. Should I specify a start action ?
[snip]
I found a better script but it wasn't multi state RA. I modified the
script
to fit my Multi State needs. I attached to this mail and you could
download
the last release (Now is the same as attached version) from:
http://code.adrianchapela.net/heartbeat/drbd_HA
Could you have a look ? Could you test it ?
Now I have solved my problem of start the drbd resource in two nodes
and
one
of them become a Master. I have now some troubles with the failover
(Old
Master is down....then Slave node detecst it but it doesn't do anything
to
be the new Master... ). I suppose an error in location rules.
Thank you
#!/bin/sh
#
# License: GNU General Public License (GPL)
# Author: Martin Fick
# Date: 04/19/07
# Origin: Hacked together from many other drbd and ocf scripts
#
# Adapted to be a MultState RA by: Adrian Chapela
# Date: 21/10/08
#
# This script manages a drbd device
#
# It can make a drbd device primary or secondary
#
# Be sure to only allow this resource to run on the
# two specific nodes where your drbd device is setup.
#
# usage: $0 {start|stop|status|monitor|meta-data|promote|demote}
#
#
# OCF parameters are as below
# OCF_RESKEY_drbd_resource
#
#######################################################################
#
#
# ocf_logi is a custom error log made by Martin Fick. I (Adrian
Chapela)
modified this function to log to another filename 'drbd_HA.log'.
# To mantain compatibility with other Hearbeat RA, I added ocf_log
in
all
locations which ocf_logi is used. This duplicates entries in logs, but
I
considered that is more
# positive to us to debug OCF script and Heartbeat configuration (to
be
it was..).
#
# We use the STATUS_CODES from here
#. /usr/lib/heartbeat/ocf-shellfuncs
# The next lines are to be more compatible with next releases of
Heartbeat
if [ -n "$OCF_DEBUG_LIBRARY" ]; then
. $OCF_DEBUG_LIBRARY
else
. ${OCF_ROOT}/resource.d/heartbeat/.ocf-shellfuncs
fi
USAGE="usage: $0 {start|stop|status|monitor|meta-data|promote|demote}";
#######################################################################
#HA_D=/etc/ha.d
#. ${HA_D}/shellfuncs
meta_data() {
cat <<END
<?xml version="1.0"?>
<!DOCTYPE resource-agent SYSTEM "ra-api-1.dtd">
<resource-agent name="drbd">
<version>0.0</version>
<longdesc lang="en">
This script manages a drbd device
It can make a drbd device primary or secondary
</longdesc>
<shortdesc lang="en">OCF MultiState Resource Agent compliant drbd
script.</shortdesc>
<parameters>
<parameter name="drbd_resource" unique="1" required="1">
<longdesc lang="en">
The drbd resource is a resource defined in /etc/drbd.conf
</longdesc>
<shortdesc lang="en">drbd resource</shortdesc>
<content type="string" default="" />
</parameter>
</parameters>
<actions>
<action name="start" timeout="1m" />
<action name="stop" timeout="1m" />
<action name="monitor" depth="10" timeout="1m" interval="5s"
start-delay="1m" />
<action name="meta-data" timeout="1m" />
<action name="promote" timeout="1m" />
<action name="demote" timeout="1m" />
</actions>
</resource-agent>
END
exit $OCF_SUCCESS
}
ocf_logi() { # type msg
# if [ "$1" != "err" ] ; then return ; fi
shift
echo `date`" - $@" >> /var/log/drbd_HA.log
}
drbd_reload() {
drbd_stop || return
drbd_start
}
drbd_stop() {
#
# Is the device already secondary?
#
# drbd_status
# if [ $? = $OCF_NOT_RUNNING ]; then exit $OCF_NOT_RUNNING; fi
#stop="$($DRBDADM secondary $RES 2>&1)"
stop="$($DRBDADM down $RES 2>&1)"
drbd_status ; rc=$?
if [ $rc = $OCF_NOT_RUNNING ]; then exit 0; fi
ocf_logi err "$RES stop failed: ($rc)"
ocf_log err "$RES stop failed: ($rc)"
ocf_logi err "$stop"
ocf_log err "$stop"
return 1
}
drbd_start() {
drbd_status
if [ $? = $OCF_SUCCESS ]; then return $OCF_SUCCESS; fi
if is_drbd_enabled; then
: OK
else
do_cmd modprobe -s drbd `$DRBDADM sh-mod-parms` || {
ocf_logi err "Can not load the drbd module."$'\n';
ocf_log err "Can not load the drbd module."$'\n';
return $OCF_ERR_GENERIC
}
ocf_logi debug "$RES start: Module loaded."
ocf_log debug "$RES start: Module loaded."
fi
if [ "$STATE" != "Secondary" ]; then
$DRBDADM up $RES
fi
# try several times, in case heartbeat deadtime
# was smaller than drbd ping time
#try=6
#while true; do
# start="$($DRBDADM primary $RES 2>&1)" && break
# let "--try" || break
# sleep 1
#done
drbd_status ; rc=$?
if [ $rc = $OCF_SUCCESS ]; then return $OCF_SUCCESS; fi
ocf_logi err "$RES start failed: ($rc)"
ocf_logi err "$start"
ocf_log err "$RES start failed: ($rc)"
ocf_log err "$start"
return $rc
}
drbd_status() {
ST=$( $DRBDADM state $RES 2>&1 )
STATE=${ST%/*}
if [ "$STATE" = "Primary" ]; then
echo "Primary - running as Master"
rc=$OCF_RUNNING_MASTER
elif [ "$STATE" = "Secondary" ]; then
echo "Secondary - Slave"
rc=$OCF_SUCCESS
else
echo "$ST"
rc=$OCF_NOT_RUNNING
fi
return $rc
}
drbd_promote() {
if is_drbd_enabled; then
: OK
else
ocf_logi err "drbd is not enabled"
ocf_log err "drbd is not enabled"
return $OCF_ERR_GENERIC
fi
drbd_status
if [ "$STATE" != "Secondary" ]; then
ocf_logi err $RES" DRBD is not prepared to be a Master in this
node ;)"
ocf_log err $RES" DRBD is not prepared to be a Master in this
node ;)"
return $OCF_ERR_GENERIC
else
if $DRBDADM primary $RES ; then
# TODO: WORK AROUND because drbdadm has a bug and
# reports success even if it failed :-(
drbd_status
if [ "$STATE" = "Primary" ]; then
ocf_logi info "$RES promote: primary succeeded"
ocf_log info "$RES promote: primary succeeded"
return $OCF_SUCCESS
else
ocf_logi err "$RES promote: Not primary despite
drbdadm call."
ocf_log err "$RES promote: Not primary despite
drbdadm
call."
fi
else
ocf_logi err "$RES promote: Failed with exit code $?."
ocf_log err "$RES promote: Failed with exit code $?."
fi
return $OCF_ERR_GENERIC
fi
}
drbd_demote() {
# Always I test drbd module
if is_drbd_enabled; then
: OK
else
ocf_logi err "drbd is not enabled"
ocf_log err "drbd is not enabled"
return $OCF_ERR_SUCCESS
fi
drbd_status
if [ "$STATE" = "Secondary" ]; then
ocf_logi err $RES" demote: already secondary"
ocf_log err $RES" demote: already secondary"
return $OCF_SUCCESS
fi
if [ "$STATE" = "Not configured" ]; then
ocf_logi debug "$RESOURCE demote: already stopped"
ocf_log debug "$RESOURCE demote: already stopped"
return $OCF_NOT_RUNNING
fi
# TODO: this is a _force_ operation. we may need to kill higher
# levels (or switch them to r/o) to be able to demote drbd.
# figure out how...
if $DRBDADM primary $RES ; then
sleep 2
drbd_status
if [ "$STATE" = "Primary" ]; then
ocf_logi err "$RESOURCE demote: still primary!"
ocf_log err "$RESOURCE demote: still primary!"
return $OCF_ERR_GENERIC
fi
ocf_logi debug "$RESOURCE demote: succeeded"
ocf_log debug "$RESOURCE demote: succeeded"
return $OCF_SUCCESS
else
ocf_logi err "$RESOURCE demote: Failed with exit code $?."
ocf_log err "$RESOURCE demote: Failed with exit code $?."
return $OCF_ERR_GENERIC
fi
return $OCF_SUCCESS
}
drbd_monitor() {
drbd_status
}
do_cmd() {
local cmd="$*"
ocf_logi Debug "$RES: Calling $cmd"
ocf_log Debug "$RES: Calling $cmd"
local cmd_out=$($cmd 2>&1)
ret=$?
if [ $ret -ne 0 ]; then
ocf_logi err "$RES: Called $cmd"
ocf_logi err "$RES: Exit code $ret"
ocf_logi err "$RES: Command output: $cmd_out"
ocf_log err "$RES: Called $cmd"
ocf_log err "$RES: Exit code $ret"
ocf_log err "$RES: Command output: $cmd_out"
else
ocf_logi debug "$RES: Exit code $ret"
ocf_logi debug "$RES: Command output: $cmd_out"
ocf_log debug "$RES: Exit code $ret"
ocf_log debug "$RES: Command output: $cmd_out"
fi
echo $cmd_out
return $ret
}
is_drbd_enabled () {
if [ -f /proc/drbd ]; then
return 0
fi
return 1
}
usage() {
echo $USAGE >&2
}
#
# Make a drbd device primary or secondary
#
DEFAULTFILE="/etc/default/drbd"
DRBDADM="/sbin/drbdadm"
if [ -f $DEFAULTFILE ]; then
. $DEFAULTFILE
fi
if
[ $# -ne 1 ]
then
usage
exit $OCF_ERR_ARGS
fi
RES="$OCF_RESKEY_drbd_resource"
case $1 in
info) cat <<-!INFO
Abstract=DRBD Device Manager
Argument=DRBD Resource Name
Description:
A DRBD device is a network raid block device which can have a
primary
and a secondary backing
device each on a separate machine. DRBD will keep them in sync.
Please rerun with the meta-data command for a list of \\
valid arguments and their defaults.
!INFO
exit $OCF_SUCCESS;;
esac
case $1 in
start|stop|status|monitor|reload|promote|demote)
ocf_logi Debug "$RES: $1"
ocf_log Debug "$RES: $1"
cmd_out=$($cmd 2>&1)
drbd_$1 ;;
meta-data) meta_data;;
usage) usage; exit $OCF_SUCCESS;;
validate-all|notify) exit 0;;
*) usage
exit $OCF_ERR_ARGS
;;
esac
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
