Bug 1709738

This bug was initially created as a copy of Bug #1707728

I am copying this bug because: 



Description of problem:
When the heavy workload as below on master, the sync is hung with sync engine tarssh. It's working fine with rsync as sync engine.

for i in {1..10000}
do
   echo "sample data" > /<mastermnt>/file$i
   mv -f /<mastermnt>/file$i /<mastermnt/file
done

Version-Release number of selected component (if applicable):
mainline

How reproducible:
Most of the times

Steps to Reproduce:
1. Setup geo-rep with master 1*3 and slave 1*3
2. touch /<mastermnt>
3. Start geo-rep and wait till the status is changelog crawl
4. Configure sync-jobs to 1
   gluster vol geo-rep <master> <slavehost>::<slavevol> config sync-jobs 1
5. Configure sync engine to tarssh
   gluster vol geo-rep <master> <slavehost>::<slavevol> config sync-method tarssh
6. Stop the geo-rep
7. Do the I/O on mastermnt as mentioned
   for i in {1..10000}
   do
      echo "sample data" > /<mastermnt>/file$i
      mv -f /<mastermnt>/file$i /<mastermnt/file
   done
8. Start the geo-rep

Actual results:
Sync hungs

Expected results:
Sync should not hung.

Additional info: