LXD 3.13 has been released

release
lxd

(Thomas Parrott) #21

OK good, I’m not going mad then. :slight_smile:

I just tested on my Ubuntu 18.04 machine with a bridged network and a 2Mbit ingress limit and speedtest-cli shows:

sudo tc class show dev $(lxc config get test1 volatile.eth0.host_name)
class htb 1:10 root prio rate 2Mbit ceil 2Mbit burst 1600b cburst 1600b 
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by CloudConnX (Eastbourne) [72.01 km]: 15.249 ms
Testing download speed................................................................................
Download: 1.58 Mbit/s
Testing upload speed................................................................................................
Upload: 18.45 Mbit/s

So it does work as far as I can tell.

What OS & kernel are you using btw?


(Thomas Parrott) #22

Can you try lowering and increasing the limits you’ve set and see if it does influence the speeds you get, even if its not near what values you’re actually setting. At least that way we can see its doing something :slight_smile:


(Saul Costa) #23

Always a good feeling :slight_smile:

It’s Ubuntu 18.04.2, kernel is 4.15.0-1032-gcp (gcp = Google Cloud Platform).

If I set them both to 1Mbit on the container level:

eth0:
  limits.egress: 1Mbit
  limits.ingress: 1Mbit
  nictype: bridged
  parent: lxdbr0
  type: nic

That does seem to do something:

Retrieving speedtest.net configuration...
Testing from Google Cloud (35.203.182.244)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by KamaTera INC (Santa Clara, CA) [11.95 km]: 24.654 ms
Testing download speed................................................................................
Download: 0.60 Mbit/s
Testing upload speed................................................................................................
Upload: 1.98 Mbit/s

The first time I tried this it actually didn’t limit the upload speed (it remained at ~20Mbit/s), but that seemed strange to me, so I tried twice more and both times it limited it (or did within a Mbit).

I also tested increasing the limits to 100Mbit each and that also worked. And then I tried my normal limits, and those also worked… gotta love these heisenbugs.

I observed similar inconsistencies when I first came across this issue. Sometimes limits appeared to work, sometimes they did not.

I’m happy to keep exploring a bit, now that things seem to be working again. I’ll run some experiments on our cluster and see if I continue to find inconsistencies.