Re: [Linux-HA] problem with resource script - possibly a timeout?

Wed, 03 Dec 2008 06:41:34 -0800

Thanks Dejan, I have it working now. I realised that the binaries were in a custom path and that it had been added to user shell profiles, including root (hence the reason the script worked when ran manually). So I added the path manually in this script and then exported. 


--
Regards

James Little

Loadbalancer.org Limited
+44 (0)870 443 8779
www.loadbalancer.org



On 3 Dec 2008, at 21:44, Dejan Muhamedagic wrote:


Hi,

On Tue, Dec 02, 2008 at 04:33:21PM +0800, James Little wrote:

Thanks for the reply Dejan. I've added code for the status action  
but still

no luck. So my script is now:

###################################################################################################
usage() {
   echo "usage: $0 start|stop"
   exit 1
}
associate() {
   CURRENT_IP=`curl -s
http://169.254.169.254/latest/meta-data/public-ipv4`

   ELASTIC_IP=`head -n 1 /etc/loadbalancer.org/aws/elasticip | tr - 
d '\n'`

   if [ "$CURRENT_IP" != "$ELASTIC_IP" ]; then
       INSTANCE_ID=`curl -s
http://169.254.169.254/latest/meta-data/instance-id`

       ec2-associate-address -i $INSTANCE_ID $ELASTIC_IP > /dev/ 
null 2>&1

& #re-associate the elastic ip with THIS instance
       exit 0


You should check the exit status or the status to make sure that
the resource has really been started. Or do you expect this never
to fail ;-)


   else #do nothing and exit, as associating twice causes problems!
       exit 0
   fi
}
disassociate() {

   ELASTIC_IP=`head -n 1 /etc/loadbalancer.org/aws/elasticip | tr - 
d '\n'`

   ec2-disassociate-address $ELASTIC_IP > /dev/null 2>&1 &

   sleep 3 #put a small delay in to make sure that this IP will  
definitely

be disassociated by the time '$0 start' runs
   exit 0


The same here: Check if the resource has been stopped.


}

get_status() {
   CURRENT_IP=`curl -s
http://169.254.169.254/latest/meta-data/public-ipv4`

   ELASTIC_IP=`head -n 1 /etc/loadbalancer.org/aws/elasticip | tr - 
d '\n'`

   if [ "$CURRENT_IP" = "$ELASTIC_IP" ]; then
       echo "Elastic IP Running OK"
       exit 0
   else
       echo "Elastic IP is stopped."
       exit 0
   fi

}

if [ $# != 1 ]; then
   usage
fi

case "$1" in
"start" ) associate;;
"stop"  ) disassociate;;
"status" ) get_status;;
*       ) usage;;
esac

####################################################################################################

here is a section of the heartbeat log:



heartbeat[23277]: 2008/12/02_03:24:47 info: Comm_now_up(): updating  
status

to active
heartbeat[23277]: 2008/12/02_03:24:47 info: Local status now set to:
'active'

heartbeat[23277]: 2008/12/02_03:24:47 WARN: No STONITH device  
configured.
heartbeat[23277]: 2008/12/02_03:24:47 WARN: Shared disks are not  
protected.
heartbeat[23277]: 2008/12/02_03:24:47 info: Resources being  
acquired from

lbslave.

harc[23286]:	2008/12/02_03:24:47 info: Running /etc/ha.d/rc.d/ 
status status
mach_down[23315]:	2008/12/02_03:24:47 info: /usr/share/heartbeat/ 
mach_down:

nice_failback: foreign resources acquired

mach_down[23315]:	2008/12/02_03:24:47 info: mach_down takeover  
complete for

node lbslave.

heartbeat[23277]: 2008/12/02_03:24:47 info: mach_down takeover  
complete.
heartbeat[23277]: 2008/12/02_03:24:47 info: Initial resource  
acquisition

complete (mach_down)

heartbeat[23287]: 2008/12/02_03:24:47 info: Local Resource  
acquisition

completed.
harc[23377]:    2008/12/02_03:24:47 info: Running
/etc/ha.d/rc.d/ip-request-resp ip-request-resp
ip-request-resp[23377]: 2008/12/02_03:24:47 received ip-request-resp
elastic OK yes

ResourceManager[23398]:	2008/12/02_03:24:47 info: Acquiring  
resource group:

lbmaster elastic
ResourceManager[23398]: 2008/12/02_03:24:47 info: Running
/etc/ha.d/resource.d/elastic  start


It looks like it is started. Perhaps try again from scratch.
Add some debugging in the script to verify that heartbeat really
invokes the start action.

Thanks,

Dejan


heartbeat[23277]: 2008/12/02_03:24:57 info: Local Resource  
acquisition

completed. (none)
heartbeat[23277]: 2008/12/02_03:24:57 info: local resource transition
completed.



As before, If I run '/etc/ha.d/resource.d/elastic  start' manually,  
then

the script works as expected. My haresources is simply:
        master elastic

where master is the local hostname.




--
Regards

James Little

Loadbalancer.org Limited
+44 (0)870 443 8779
www.loadbalancer.org



On 1 Dec 2008, at 21:19, Dejan Muhamedagic wrote:


Hi,

On Mon, Dec 01, 2008 at 01:47:56PM +0800, James Little wrote:

Hi all,


I'm currently working with heartbeat in the Amazon cloud and (for  
those

that are familiar with Amazon EC2), I'm trying to use heartbeat to
re-allocate an elastic IP on detection of node failure. I'm not
experienced

with writing my own heartbeat resource scripts, but this is what  
I have:


usage() {
  echo "usage: $0 start|stop"
  exit 1
}
associate() {
  CURRENT_IP=`curl -s
http://169.254.169.254/latest/meta-data/public-ipv4`
  ELASTIC_IP=`head -n 1 /etc/loadbalancer.org/aws/elasticip | tr -d
'\n'`
  if [ "$CURRENT_IP" != "$ELASTIC_IP" ]; then
      INSTANCE_ID=`curl -s
http://169.254.169.254/latest/meta-data/instance-id`

      ec2-associate-address -i $INSTANCE_ID $ELASTIC_IP > /dev/ 
null 2>&1

& #re-associate the elastic ip with THIS instance
      exit 0
  else #do nothing and exit, as associating twice causes problems!
      exit 0
  fi
}
disassociate() {
  ELASTIC_IP=`head -n 1 /etc/loadbalancer.org/aws/elasticip | tr -d
'\n'`
  ec2-disassociate-address $ELASTIC_IP > /dev/null 2>&1 &
  sleep 3 #put a small delay in to make sure that this IP will
definitely
be disassociated by the time '$0 start' runs
  exit 0
}

if [ $# != 1 ]; then
  usage
fi

case "$1" in
"start" ) associate;;
"stop"  ) disassociate;;
*       ) usage;;
esac



The RA must support the status action. You already have code for
it in the start action. See:

http://www.linux-ha.org/HeartbeatResourceAgent

Thanks,

Dejan



and I've simply put this script in /etc/ha.d/resource.d. The script

definitely runs (via heartbeat) and I have put some debug stuff  
in to

confirm this (piping output to files etc.) but for some reason the
ec2-associate-address and ec2-disassociate-address commands have no
effect,
but when tested manually these commands work ok. In fact this whole
script,

when ran manually, behaves as expected. The heartbeat log  
indicates that

the script has ran.


I'm running heartbeat v2.1.3, but have not enabled CRM. I did  
think it

may

be a timeout because the ec2 commands take a few seconds to run,  
but I

tested a trivial resource script that did 'sleep 10; echo 'stuff' >

test-file' and that ran without problems, despite the 10 second  
sleep.



Any help greatly appreciated; thanks in advance...



--
Regards

James Little

Loadbalancer.org Limited
+44 (0)870 443 8779
www.loadbalancer.org



_______________________________________________
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


_______________________________________________
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


_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems






















					Reply via email to