OpenNebula

Hi

Please note that the pre/post scripts are from the system datastore, more info here¹. Could you check the TM_MAD of the system datastore, and check that those scripts are not called...

[1] http://docs.opennebula.org/4.14/integration/infrastructure_integration/sd.html#tm-drivers-structure

Thanks Ruben. Yes it seems to be that case. But I do not understand how it could be managed. If the VM has attached disk from other then system datastore, it meens it cannot be live migrated? Or how could I call datastore TM scripts from system datastore TM scripts?

Hi Michal, The shared system ssh, that can work, with multiple image datastores already have this:

#!/bin/bash

DRIVER_PATH=$(dirname $0)

DISK_COUNT=$(onevm show $4 -x|grep DISK_ID| wc -l)
TMS=$(onevm show $4 -x|sed -rn 's/[[:space:]]*<TM_MAD><\!\[CDATA\[([^]]*).*/\1/p')

XPATH="${DRIVER_PATH}/../../datastore/xpath.rb -b $7" 

for i in `seq 1 $DISK_COUNT`; do
  TM=`echo $TMS|cut -d" " -f$i`
  DISK_ID=$(($i-1))
  DEV=`ssh $1 "readlink $3/disk.$DISK_ID"`
  ${DRIVER_PATH}/../$TMS/premigrate "$1" "$2" "$DEV" 
done

exit 0

Maybe you can need to use the same logic in ssh? or this is not working for you?

Yes Ruben it seems it solves my problem. This code is not from 4.14 so I suppose it is from upcomming 5.0. So I will just wait for 5.0 :-)