Score:0

WonderShaper limits the bandwidth too much

fr flag

We use WonderShaper as a traffic control tool on our linux servers. We have 10Gbit uplink, and I want to limit the upload and download to be up to 1Gbit per server. My goal is to prevent a situation in which one server downloads a huge file and chokes the whole line.

All I need to do is to define what is my required upload and download bandwidth (kb) -

wondershaper -a enp0s3 -d 1024000 -u 1024000

However, whenever I'm testing iperf from a client to the same server I made the changes on - iperf3 -c {IPERF_SERVER} -t 60 I'm getting very low results, much less than the configured:

[ ID] Interval           Transfer     Bandwidth       Retr
[  4]   0.00-100.00 sec  3.42 GBytes   294 Mbits/sec  2006             sender
[  4]   0.00-100.00 sec  3.41 GBytes   293 Mbits/sec                  receiver

Testing download (adding -R to iperf) leading to 950Mbit. Whenever I'm disabling WonderShaper, I'm getting almost 10Gbit for both upload and download.

Does anybody familiar with these issues using WonderShaper? Or is there anything in addition I can do in order to solve it?

That's the output of wondershaper -s -a enp0s3

qdisc htb 1: root refcnt 2 r2q 10 default 32 direct_packets_stat 0 direct_qlen 1000
 Sent 42101739 bytes 592924 pkt (dropped 0, overlimits 133 requeues 0)
 backlog 0b 0p requeues 0
qdisc sfq 30: parent 1:30 limit 127p quantum 6000b depth 127 divisor 1024 perturb 10sec
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
qdisc sfq 10: parent 1:10 limit 127p quantum 6000b depth 127 divisor 1024 perturb 10sec
 Sent 37916360 bytes 574353 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
qdisc sfq 20: parent 1:20 limit 127p quantum 6000b depth 127 divisor 1024 perturb 10sec
 Sent 4185379 bytes 18571 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
qdisc ingress ffff: parent ffff:fff1 ----------------
 Sent 3717886081 bytes 533558 pkt (dropped 639, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
class htb 1:1 root rate 1024Mbit ceil 1024Mbit burst 1408b cburst 1408b
 Sent 42102725 bytes 592931 pkt (dropped 0, overlimits 58 requeues 0)
 backlog 0b 0p requeues 0
 lended: 38 borrowed: 0 giants: 0
 tokens: 163 ctokens: 163

class htb 1:10 parent 1:1 leaf 10: prio rate 204800Kbit ceil 1003Mbit burst 1561b cburst 1379b
 Sent 37917346 bytes 574360 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
 lended: 574360 borrowed: 0 giants: 0
 tokens: 852 ctokens: 163

class htb 1:20 parent 1:1 leaf 20: prio rate 409600Kbit ceil 972800Kbit burst 1536b cburst 1459b
 Sent 4185379 bytes 18571 pkt (dropped 0, overlimits 77 requeues 0)
 backlog 0b 0p requeues 0
 lended: 18360 borrowed: 38 giants: 0
 tokens: 225 ctokens: 94

class htb 1:30 parent 1:1 leaf 30: prio rate 204800Kbit ceil 921600Kbit burst 1561b cburst 1382b
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
 lended: 0 borrowed: 0 giants: 0
 tokens: 968 ctokens: 203
mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.