Commit b8f68588 authored by David Johnson's avatar David Johnson

Put back the old non-parallelized nova boot path; option for future.

I won't get the Neutron lock bug solved in time for this eval churn, so
for now, undo the path where we launch all the slaves via a single 'nova
boot' call.  This caused lockups in Neutron after N slaves had been
created; cause unknown.

Doing a call per VM probably isn't much slower in the long run, anyway.
parent 744cf6d7
Pipeline #1888 passed with stage
in 2 seconds
......@@ -30,6 +30,12 @@ if [ $# -lt 6 ]; then
exit 1
fi
#
# Launching all slaves from one nova command causes a Neutron lock!
# So don't do this for now by default.
#
FASTBOOT=0
TESTDIR=$1
shift
UTENANT=$1
......@@ -108,10 +114,22 @@ if [ ! $? -eq 0 ]; then
echo "ERROR: failed to create resourcemanager VM; aborting!"
exit 1
fi
nova boot --image hadoop --flavor m1.medium --nic net-id=$NETWORKID --min-count $NSLAVES slave
if [ ! $? -eq 0 ]; then
echo "ERROR: failed to create slave VMs; aborting!"
exit 1
if [ $FASTBOOT -eq 1 ]; then
nova boot --image hadoop --flavor m1.medium --nic net-id=$NETWORKID --min-count $NSLAVES slave
if [ ! $? -eq 0 ]; then
echo "ERROR: failed to create slave VMs; aborting!"
exit 1
fi
else
i=0
while [ $i -lt $NSLAVES ]; do
nova boot --image hadoop --flavor m1.medium --nic net-id=$NETWORKID slave-$i
if [ ! $? -eq 0 ]; then
print "ERROR: failed to create slave-$i VM; aborting!"
exit 1
fi
i=`expr $i + 1`
done
fi
#
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment