We transfer several terabytes of virtual servers to a remote site (5-6 at the same time due to the latency to the remote site).
Out of the 5-6 sometimes 1 or 2 fail, and we get these message in the log file on the server side:
2012/03/28 13:42:12 rsync: writefd_unbuffered failed to write 4092 bytes to socket : Connection reset by peer (104)
2012/03/28 15:06:41 rsync: writefd_unbuffered failed to write 4092 bytes to socket : Connection reset by peer (104)
On the client side rsync.exe gets stuck on
sending incremental file list
The connection between the sites, although 100ms, is extremely stable. I have a ping going the whole time between the two nodes and there is almost zero packet loss. Our command line for the client is as follows:
c:\cwrsync\bin\rsync --timeout=1800 --block-size=32768 -v -r --inplace --ignore-times --progress --include='*/' --include='*.vhd*' --include='*.VHD*' --exclude='*' /cygdrive/v/%%i rsync://xx.xx.xx.xx/HPV
(ip address is anonymized by tk)
The VHD files are quite large, ranging from 40GB to 150GB, but the failure is not size dependent, it happens to all of them.
Any ideas how to fix/work around this?