* Re: [Bloat] link saturation test failing
@ 2019-01-03 17:16 Dev
2019-01-03 17:18 ` Dave Taht
0 siblings, 1 reply; 5+ messages in thread
From: Dev @ 2019-01-03 17:16 UTC (permalink / raw)
To: bloat
Dev <dev@logicalwebhost.com> writes:
> I’m running Debian Buster, trying to saturate a link, enable fq_codel
> and drop my ping times back down, but they seem to stay high.
>
> I connected machine A and B to a gig switch, then dropped the port
> speed on machine A like:
>
> ethtool -s enp8s0 advertise 0x002
> From which machine are you running the test? If your bottleneck is the
> switch, that is where the buffering will happen; FQ-CoDel on the host
> can do nothing about that...
> -Toke
I’m running the test from machine A, my Debian Buster server, so that’s the interface I’ve set to 10Mbps. Basically trying to follow the howto on the website, but my ping times aren’t dropping once fq_codel is enabled.
-Dev
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Bloat] link saturation test failing
2019-01-03 17:16 [Bloat] link saturation test failing Dev
@ 2019-01-03 17:18 ` Dave Taht
2019-01-03 21:17 ` Dev
0 siblings, 1 reply; 5+ messages in thread
From: Dave Taht @ 2019-01-03 17:18 UTC (permalink / raw)
To: Dev; +Cc: bloat
My guess is your ethernet device does not have BQL. To check, cd
/sys/class/net/your_device/queues/tx-0/byte_queue_limits and see if
you have a limit other than 0.
On Thu, Jan 3, 2019 at 9:16 AM Dev <dev@logicalwebhost.com> wrote:
>
> Dev <dev@logicalwebhost.com> writes:
>
> > I’m running Debian Buster, trying to saturate a link, enable fq_codel
> > and drop my ping times back down, but they seem to stay high.
> >
> > I connected machine A and B to a gig switch, then dropped the port
> > speed on machine A like:
> >
> > ethtool -s enp8s0 advertise 0x002
>
> > From which machine are you running the test? If your bottleneck is the
> > switch, that is where the buffering will happen; FQ-CoDel on the host
> > can do nothing about that...
>
> > -Toke
>
> I’m running the test from machine A, my Debian Buster server, so that’s the interface I’ve set to 10Mbps. Basically trying to follow the howto on the website, but my ping times aren’t dropping once fq_codel is enabled.
>
> -Dev
> _______________________________________________
> Bloat mailing list
> Bloat@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/bloat
--
Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-205-9740
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Bloat] link saturation test failing
2019-01-03 17:18 ` Dave Taht
@ 2019-01-03 21:17 ` Dev
0 siblings, 0 replies; 5+ messages in thread
From: Dev @ 2019-01-03 21:17 UTC (permalink / raw)
To: Dave Taht; +Cc: bloat
#>: cd /sys/class/net/eth0/queues/tx-0/byte_queue_limits
/sys/class/net/eth0/queues/tx-0/byte_queue_limits >: ls
hold_time inflight limit limit_max limit_min
/sys/class/net/eth0/queues/tx-0/byte_queue_limits >: cat limit
250
- Dev
> On Jan 3, 2019, at 9:18 AM, Dave Taht <dave.taht@gmail.com> wrote:
>
> My guess is your ethernet device does not have BQL. To check, cd
> /sys/class/net/your_device/queues/tx-0/byte_queue_limits and see if
> you have a limit other than 0.
>
> On Thu, Jan 3, 2019 at 9:16 AM Dev <dev@logicalwebhost.com> wrote:
>>
>> Dev <dev@logicalwebhost.com> writes:
>>
>>> I’m running Debian Buster, trying to saturate a link, enable fq_codel
>>> and drop my ping times back down, but they seem to stay high.
>>>
>>> I connected machine A and B to a gig switch, then dropped the port
>>> speed on machine A like:
>>>
>>> ethtool -s enp8s0 advertise 0x002
>>
>>> From which machine are you running the test? If your bottleneck is the
>>> switch, that is where the buffering will happen; FQ-CoDel on the host
>>> can do nothing about that...
>>
>>> -Toke
>>
>> I’m running the test from machine A, my Debian Buster server, so that’s the interface I’ve set to 10Mbps. Basically trying to follow the howto on the website, but my ping times aren’t dropping once fq_codel is enabled.
>>
>> -Dev
>> _______________________________________________
>> Bloat mailing list
>> Bloat@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/bloat
>
>
>
> --
>
> Dave Täht
> CTO, TekLibre, LLC
> http://www.teklibre.com
> Tel: 1-831-205-9740
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [Bloat] link saturation test failing
2018-12-27 17:29 Dev
@ 2018-12-31 20:39 ` Toke Høiland-Jørgensen
0 siblings, 0 replies; 5+ messages in thread
From: Toke Høiland-Jørgensen @ 2018-12-31 20:39 UTC (permalink / raw)
To: Dev, bloat
Dev <dev@logicalwebhost.com> writes:
> I’m running Debian Buster, trying to saturate a link, enable fq_codel
> and drop my ping times back down, but they seem to stay high.
>
> I connected machine A and B to a gig switch, then dropped the port
> speed on machine A like:
>
> ethtool -s enp8s0 advertise 0x002
From which machine are you running the test? If your bottleneck is the
switch, that is where the buffering will happen; FQ-CoDel on the host
can do nothing about that...
-Toke
^ permalink raw reply [flat|nested] 5+ messages in thread
* [Bloat] link saturation test failing
@ 2018-12-27 17:29 Dev
2018-12-31 20:39 ` Toke Høiland-Jørgensen
0 siblings, 1 reply; 5+ messages in thread
From: Dev @ 2018-12-27 17:29 UTC (permalink / raw)
To: bloat
I’m running Debian Buster, trying to saturate a link, enable fq_codel and drop my ping times back down, but they seem to stay high.
I connected machine A and B to a gig switch, then dropped the port speed on machine A like:
ethtool -s enp8s0 advertise 0x002
Then verified the link lights on switch show 10M. Then try to enable fq_codel and swamp link like:
tc qdisc add dev enp8s0 root fq_codel
dd if=/dev/urandom | ssh user@192.168.10.150 dd of=/dev/null
I still have high ping times (10msec vs 0.5 msec). I try to saturate with iperf and the same thing happens. Then I try to look at fq_codel like:
tc qdisc show dev enp8s0
qdisc fq_codel 8001: root refcnt 6 limit 10240p flows 1024 quantum 1514 target 5.0ms interval 100.0ms memory_limit 32Mb ecn
and
tc -g -s class show dev enp8s0
+---(:5) mq
| Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
| backlog 0b 0p requeues 0
|
+---(:4) mq
| Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
| backlog 0b 0p requeues 0
|
+---(:3) mq
| Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
| backlog 0b 0p requeues 0
|
+---(:2) mq
| Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
| backlog 0b 0p requeues 0
|
+---(:1) mq
Sent 248411320 bytes 164594 pkt (dropped 0, overlimits 0 requeues 4)
backlog 0b 0p requeues 4
But I don’t really know what I should be looking for/doing.
-Dev
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2019-01-03 21:17 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-01-03 17:16 [Bloat] link saturation test failing Dev
2019-01-03 17:18 ` Dave Taht
2019-01-03 21:17 ` Dev
-- strict thread matches above, loose matches on Subject: below --
2018-12-27 17:29 Dev
2018-12-31 20:39 ` Toke Høiland-Jørgensen
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox