I tried to find sth. like a process bar, but failed. Thanks for any info.
In the logs you will find a line similar to:
migrates in progress ( 0 , 0 ) ::: ClusterSize 36 ::: objects 200435832
Which by default is printed every 10 seconds. This line is described here. Basically you would want to monitor the first first number in the tuple which is the total number of migrations this node expects to send. You could then sum this value across the cluster and monitor some time to determine the rate at which this value is decreasing.
This value can also be retrieved with the following info command:
asinfo -v "statistics" -l | grep migrate_progress_send
But there is a problem with this method, in my monitoring experience, there seems to be two scheduling rounds for migrations so this would give you an ETA until the current round is completed. Normally the second round is smaller than the first though.
Thanks. But it seems not accurate, if just one node reboots. At first, all the other nodes will finish sending, then its sending will last a relative longer time, and it is also my observation.