Hi there

I wish to control bandwidth to a particular IP address for their downloads on a slow connection to shape how much it can use at a given time. So far I've tried Luci QOS and SQM and it has made zero effect on the device using all the download bandwidth.

I do know that gargoyle is a distribution that supports this, but it seems odd that if that does, open wrt doesn't....

I would like to try this http://cmetcalfe.ca/blog/limit-download … dress.html however if possible I'd like to make a Luci app for all to use for this purpose, my old router supported this and I found it very useful! (Bandwidth control by IP Address...

I would also like to have wrtbrmon webpage accessible in Luci, is there a quick a d easy way to add a www shortcut to Luci?

So far all those requesting such a feature either happily went the tomato/gargoyle way or just disappeared/despaired. Unfortunately before fully diagnosing what exactly was/is causing the failure of qos-scripts/sqm-scripts to deal sufficiently with the culprit. Now such an analysis will not guarantee that either can be adjusted/taught to deal with the offending traffic pattern, but without sch analysis/data, there is not much probability of anything happening at all.
        My take on why openwrt is not equal to gargoyle is that then one of them would be superfluous Or rather that for most people qos/sqm work well enough.
        The challenge I see with the linked approach is that ideally one configures a shaper to manage the whole bandwidth of a link, because otherwise it gets hard to guarantee the shaped IP any traffic at all (which might be the requested policy, but still it would be nice if there was choice). And for most routers the maximum bandwidths that can be shaped is in the 50-100Mbps up- and downstream shaping combined (newer models might reach 200Mbps).

Now with SQM scrips you can simply use the available machinery ans modify say /usr/lib/sqm/simple.qos and just add a 4th HTB tier and an iptables filter to direct packets from and to $IP to that tier...

Best Regards
       M.

(Last edited by moeller0 on 22 May 2016, 16:17)

Hello,I heard that you are a guru.I am struggling with my strange ISP.Any idea you can share?
https://forum.openwrt.org/viewtopic.php?pid=325603

(Last edited by axishero on 24 May 2016, 11:23)

Hi axishero,

I feel flattered, but alas I am no guru in any shape or form. I can offer to help you help yourself if you want to spend the time though. I would need more detail about your setup and constraints from your ISP and you would need to test everything (since I can not recreate your specific circumstances due to a lack of time and hardware), so if we embark upon this, you would need to do all the "heavy lifting" while I "cheer/coach" from the side lines. Also I am a bit short on time so any incremtal steps would be slow (as long as you need my input that is). And no, waiving bitcoins my way is not going to change that, I have a day job (and more importantly a family) so my priorities are elsewhere. That said, I am happy to help as I will learn something new in it... BTW, the delayed response to your post is predictive of how timely my input would be...

Best Regards
        M.

network wise I cant see the issue, I can easily do it within CISCO routers at work, albeit they are thousands of dollars worth. However I used to get this functionality from y old router which is still Linux based, perhaps ill have to decompile it to see how it works

I don't know if this documentation is relevant anymore?  https://wiki.openwrt.org/doc/howto/pack … .scheduler

So there are basically two challenges as I see it. The first one is rather simple all one needs to do it actually sit down and implement it. The second is more severe, typically bandwidth shapers are instantiated at the WAN interface (as for a typical home router if you instantiate it on a LAN interface the WLAN/WIFI traffic is not shaped). But on the WAN interface we only see the NAT'ed internal addresses, on egress I believe the conntrack system would allow us to see the internal src IP address of outgoing packets, but on ingress using an IFB we do not have access to the netfilter system yet, and have to relay on tc. And as far as I know tc has no way of getting at the internal dst IP address of incoming packets. So there it is, the kind of shaping that people seem to want is simply hard to implement for the typical wifi-routers pople use at home...

Best Regards
        M.