[GitHub] [incubator-pegasus] levy5307 commented on a change in pull request #576: fix(script): operation return before the cluster balance when using pegasus_rolling_update.sh

[GitBox <gi...@apache.org>]

levy5307 commented on a change in pull request #576:
URL: https://github.com/apache/incubator-pegasus/pull/576#discussion_r472868001



##########
File path: scripts/pegasus_rebalance_cluster.sh
##########
@@ -0,0 +1,121 @@
+#!/bin/bash
+#
+# Pegasus cluster rebalance 
+#
+
+PID=$$
+
+if [ $# -le 2 ]; then
+  echo "USAGE: $0 <cluster-name> <cluster-meta-list> <only-move-primary>"
+  echo
+  echo "for example:"
+  echo "  $0 onebox 127.0.0.1:34601,127.0.0.1:34602 true"
+  echo
+  exit 1
+fi
+
+cluster=$1
+meta_list=$2
+only_move_primary=$3
+
+pwd="$( cd "$( dirname "$0"  )" && pwd )"
+shell_dir="$( cd $pwd/.. && pwd )"
+cd $shell_dir
+
+source ./scripts/minos_common.sh
+find_cluster $cluster
+if [ $? -ne 0 ]; then
+  echo "ERROR: cluster \"$cluster\" not found"
+  exit 1
+fi
+
+echo "UID=$UID"
+echo "PID=$PID"
+echo "Start time: `date`"
+rebalance_start_time=$((`date +%s`))
+echo
+
+echo "Generating /tmp/$UID.$PID.pegasus.rebalance.cluster_info..."
+echo cluster_info | ./run.sh shell --cluster $meta_list 2>&1 | sed 's/ *$//' >/tmp/$UID.$PID.pegasus.rebalance.cluster_info
+cname=`grep zookeeper_root /tmp/$UID.$PID.pegasus.rebalance.cluster_info | grep -o '/[^/]*$' | grep -o '[^/]*$'`
+if [ "$cname" != "$cluster" ]; then
+  echo "ERROR: cluster name and meta list not matched"
+  exit 1
+fi
+pmeta=`grep primary_meta_server /tmp/$UID.$PID.pegasus.rebalance.cluster_info | grep -o '[0-9.:]*$'`
+if [ "$pmeta" == "" ]; then
+  echo "ERROR: extract primary_meta_server by shell failed"
+  exit 1
+fi
+
+if [ "$only_move_primary" == "true" ]; then
+  echo "Set meta.lb.only_move_primary true"
+  echo "This remote-command tells the meta-server to ignore copying primaries during rebalancing."
+  echo "So the following steps only include move_primary and copy_secondary."
+  echo "remote_command -l $pmeta meta.lb.only_move_primary true" | ./run.sh shell --cluster $meta_list &>/tmp/$UID.$PID.pegasus.rebalance.only_move_primary
+  set_ok=`grep OK /tmp/$UID.$PID.pegasus.rebalance.only_move_primary | wc -l`
+  if [ $set_ok -ne 1 ]; then
+    echo "ERROR: meta.lb.only_move_primary true"
+    exit 1
+  fi
+fi
+echo
+
+echo "Set meta level to lively..."
+echo "set_meta_level lively" | ./run.sh shell --cluster $meta_list &>/tmp/$UID.$PID.pegasus.rebalance.set_meta_level
+set_ok=`grep 'control meta level ok' /tmp/$UID.$PID.pegasus.rebalance.set_meta_level | wc -l`
+if [ $set_ok -ne 1 ]; then
+  echo "ERROR: set meta level to lively failed"
+  exit 1
+fi
+
+echo "Wait cluster to become balanced..."
+echo "Wait for 3 minutes to do load balance..."
+sleep 180
+op_count_check_time=1

Review comment:
       I think `op_count_check_remain_times` is better




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@pegasus.apache.org
For additional commands, e-mail: dev-help@pegasus.apache.org