Bandwidth graph: Transmit and Receive swapped?
-
It looks like the transmit and receive rates are swapped in the bandwidth graph:
All of that bandwidth is due to
FOGImageReplicator
copying my images fromlab2-server
toGleeble
, but the graph (as far as I can see) shows thatGleeble
is the one transmitting lots of data.Fog version ??? (SVN 5892) is running on Ubuntu 14.04 on lab2-server, which is also the master node.
-
The values should not be flipped any more. I mean worse things have happened, and I suppose it’s possible Gleeble (eth0) and lab2-server (eth0) are on the same setup.
-
Hi Tom, what do you mean by “on the same setup”? These are two physically distinct boxes in two different rooms.
-
@dolf I was only guessing.
If they had been on the same setup, it likely would’ve looked a lot stranger (like both nodes would’ve had equal bandwidth usage (slightly delayed because they have to check in two different urls).
-
Please update, at the least the version will be fixed.
-
When I
scp
fromlab2-server
toGleeble
, it still shows an orange line instead of a blue one. When I copy something fromgleeble
to another PC, it shows thatlab2-server
is transmitting (the blue spike):Maybe toe transmit/receive is not swapped, but rather the server names?
-
@dolf I found this to be a bug, read here: https://forums.fogproject.org/topic/8076/bandwidth-chart-and-location-plugin
-
Great! Sorry for the re-post then.
-
I’m hoping a newer commit has addressed this. I’m not “overly” concerned with the graph as all it does is give you an overview. Basically, yes I want to fix, no I don’t know why it’s coming back reversed considering the “failsafe’s” are already in place and should be operational. I hope by using ksort rather than asort that this would now be fixed.
-
Yep, it works now.
Just for fun: This is the command I use to pull random data from a remote host (and monitor the speed):
$ ssh user@remote "cat /dev/urandom" | pv > /dev/null
-
@dolf That is a pretty sweet command. However, have you heard of iPerf? Or even Flent?
-
Nope. I’ll check it out. But piping stuff around in linux is simple and effective.
-
@dolf Simple doesn’t get the job done when you’re trying to measure bufferbloat, or other multi-variable problems
Check out Flent: https://flent.org/