@jacquesh said in internet slow upload:

the problem is the combination of pfsense and setting up the ppoe connection.

we have a couple of APU4d4 boxes configured with PPPoE, it runs smoothly...

@Gertjan "his idea is perfect to find out where is the bottleneck might be..
we use an old UBNT EdgeRouterX (penny stuff) for PPPoE test, CPU dual-Core 880Mhz, so it is especially suitable for testing

@viktor_g The patch worked perfectly!

Well, in todays virtualized world, it would be possible to run two instances of pf, one doing ids/ips functions and then feed it to the second instance to do traffic shaping. Just an idea.

@finnschi - I'm not sure 48ms is enough added latency to see adverse effects. If you bump that up to 150-200ms I think it will start to become more noticeable. Also, regarding bandwidth - modern remote desktop/screen sharing programs are pretty efficient these days and don't need a significant amount of bandwidth. What did your additional testing reveal?

@markn6262 May i ask for what use case do you need evenly shared bandwidth between IP's?
The solution you mentions with Codel/WFQ2+ doesn't care about the performance of individual applications, with that i mean every single IP is a queue/flow with that solution.
With fq-codel every connection established gets it's own queue/flow, depending on the limit of flows set in settings, which makes it able to prioritize latency sensitive traffic no matter what host it comes from and make sure that no host can starve the connection for bandwidth.
The point i am getting at is that trying to share bandwidth evenly between IP addresses doesn't guarantee optimal performance for low latency applications, fq-codel tries to, so when you are having several host trying to do large downloads they still get priority for other applications like VoIP, gaming or just streaming netflix if using fq-codel. The thing your going for was smart 15 years ago but only because there was nothing better.

@viktor_g Mbit default on "New Limiter" now, much appreciated.

Removing/renaming limiters fixed in upstream:
https://redmine.pfsense.org/issues/3924

@discy Fixed in upstream: https://redmine.pfsense.org/issues/10660#note-8

@YossiZa said in Hi CPU usage when have high traffic:

E5-2680 v2

E5-2680 v2 has 40% more memory bandwidth. CPU usage is not just a metric of how hard a CPU is working, it's a metric of how many cycles the CPU is having to wait for work to be done. CPU bound work is as fast as the CPU can process it. But memory bound work has to wait for the memory to respond. If the memory is at its max, then the CPU will spend more time waiting not doing anything. The more cache can't hurt either. Fewer memory accesses required.

Thanks. Couldn't find a definitive answer to if and when the interface method could be usefull.

Unfortunately my question remains unanswered :).

We haven't got around to digging into that one yet, lots of other higher priorities.

@remlei said in Inaccessible pfsese web admin page when traffic shaping is enabled:

internet speed to compensate the congestion with the bandwidth.

You looking to mitigate the bufferbloat on WAN ?
http://www.dslreports.com/speedtest

found this

Bandwidth (root queues)
The amount of bandwidth available on this interface in the outbound direction. For example, WAN-type interface root queues list upload speed. LAN-type interfaces list the sum total of all WAN interface download bandwidth.

https://docs.netgate.com/pfsense/en/latest/book/trafficshaper/advanced-customization.html

so set 32 Mbit on the WAN interface and 50mbit on LAN Interfaces.

These settings are not limiting the max bandwith in case I would get more bandwith, just used for the algorithm priorizing the packets, right?