volrep of 2TB block lv running at 8+MB/s?

[pwason]

Maybe I'm doing bad math, but this is what's going on:

I have a 2TB block-i/o volume that's about 54% full with VM disk images (all the VMs are shutdown). I'm using synchronous volrep to copy them to an identical volume on another storage box. The volrep task has been running for about 4.5 hours at around 65Mbps. That gives me about 125GB copied so far. Which means the whole thing is going to take close to 40 hours!

I think this is running so slow because I left the default bandwidth setting when I created the volrep task. I can basically use as much bandwidth as I want - there's only one other user logged in and they're not even doing anything on the network. Should I stop this task, delete it, clear destination metadata, and start a new volrep task with a much higher bandwidth? Can I just peg it at 1024?

[To-M]

It shouldnt take that long, so there could be some network issue. Kill the task then recreate the task, if 1GbE NIC then use 100 if 10GbE (prefered) use 600.

[pwason]

(even though it's on 1GbE..) reduced time to ~3.5 hrs, which was about what I expected when I started all of this.

Still was only about 15MB/sec, at around 80k packets/sec?

TFTR, Todd

[To-M]

On a 1GbE you can also try with increasing the jumbo frames in the Console screen CTRL + ALT + W then select the Tuning options, but I would let this run first.