aMule Forum
English => en_Bugs => Topic started by: stoatwblr on April 16, 2012, 04:18:27 PM
-
Using upnp-router-control, I can see a large inbound traffic spike (nearly 2 times normal flow) every 11 seconds (100kB/s will spike to 180, 200 to 350, etc)
Outbound traffic spikes are smaller but happen about the same frequency.
Neither peak is enough to hit througput limits or ISP throttle values.
This only happens when amule(d) is running (tested on multiple machines), so I'm sure it's not caused by anything else.
Does anyone have any idea what these spikes are?
-
Do you have a download speed limit? Does it go away if you remove it? WX or Asio?
-
Do you have a download speed limit? Does it go away if you remove it? WX or Asio?
The limit is my wire speed and it stays there with limits removed or set lower, just varies in amplitude.
WX and asio, multiple versions.
I noticed it a few months ago, but didn't start timing it and looking for reasons until the last few days - that's when I found the papers on ptp detection based on heartbeat intervals.
-
Ok, I wanted to know if the bandwidth limiting code is involved.
Wait... You regularly download at limit of your line speed? That's very unusual. UL speed is near const, DL varies greatly.
How did you measure these spike speeds? How would that even be possible if you are already at your line speed?
-
Ok, I wanted to know if the bandwidth limiting code is involved.
No, it happens if it's disabled.
Wait... You regularly download at limit of your line speed?
Hell no. That's 17Mb/s (which I can achieve weasily on FTP and HTTP transfers from $orkplace)
The usual speed for Amule is 1-3Mb/s. The peaks are 50-100% higher - but this is still well below line speed and well below what I have the limiter set to (about 15Mb/sec)
How did you measure these spike speeds? How would that even be possible if you are already at your line speed?
upnp-router-control samples and graphs traffic every second, giving the last 90 seconds. This complements using mrtg for longer trends (I really should move to cactii, but fritzboxes don't do snmp)
-
Is there just the spike up, or is there a spike down before the spike too? (Like not receiving for a little while and then taking all that's arrived.)
Something matching in the CPU load?
-
No, nothing like that. The spikes are above a steady baseline.