Could someone please run iperf3 as a client from a router? I experience strange behavior - throughout is 10-15 mbits. When iperf3 is ran as a server on r7800, then throughput is 850-900 mbits (guess it's my pc nics top)
If someone is interested I've made a startup script that adjusts smp affinity.
1 Like
@Nague: Are you sure IGMP snooping worked correctly also for IGMPv2? My testing indicates that it does not, and I am really frustrated at the moment. I made a post about it here, but the rundown is that with IGMP snooping enabled, the switch drops all IGMP query messages.
What this means, is that a client may send a report/join message to a server, and start receiving the stream, but when the server (after some timeout) sends a query to see if the client still wants to receive the stream, the query is dropped on ingress, so the client never gets it. Thus the server never gets a response to the query, and stops the stream.
Immensely annoying!
@mroek : I'm not sure IGMP snooping feature works correctly in v2. We can try to troubleshoot, let's continue in your post.
@dissent1: I've been thinking about testing a little bit with the new DSA driver, but I'm having a hard time finding proper documentation as to how I'd set whether a port is a tagged or untagged member of a VLAN, and also how to set the default VLAN? Do you use VLANs in your setup? The old switch UCI config is not adjusted for this driver, is it? If it is, then this question is irrelevant, of course. When it comes to IGMP snooping, it seems that the only setup for that is for a bridge, so if IGMP snooping is set for br-lan, then the driver should automatically set the proper switch config for that?
Here's how you set vlans, screenshot of br-lan -> physical settings
Yes that's correct, in theory. By design the driver propagates kernel settings into the switch, so it behaves like each port is a real NIC. But as I get it it's more like multicast offloading, so multicast forwarding is done within the switch.
1 Like
I'm not entirely clear what is set in that screenshot. All the ticked interfaces are bridged, and the custom interface belongs to VLAN 3 on port lan1? So incoming packets on lan 1 with VLAN id 3 is then forwarded (untagged) within the bridge? What about egress? How do you set whether a interface is a tagged or an untagged member of a VLAN?
On that screenshot - lan port 1 is tagged with vid 3, lan 2-4 are untagged, you can combine whatever you want.
A port cant be untagged member of vlan - it's wether a member of vlan and tagged or untagged completely there's no 3rd option. What you have seen with luci switch before where you can set cpu and port tagged independently - is somewhat excessive configuration, or misconfiguration if you call it.
If you want to tag a port you add .vid to it i.e. lan1.3 or 'wan.8' etc. you can add as many vids to a single port as you want and bridge em together, route or send to the moon (not really).
Yes, exactly, and they all are bridged together - can communicate with each other
All the packets that are destined to lan port 1 in that configuration are tagged with vid 3, that's pretty simple 
edit: forgot to mention - after you fill custom interface field with custom interface you it apply and it's added to the list above it
@philjohn it's mostly fixes there that I've gathered from qsdk, Shortcut fe and qca8k driver (stripped version from blogic's repo).
Ok, thanks. I think I got it. I'll play around a little with it in the near future, but since the wifi is unstable and useless (for me) in master, I'll need to run with stable for daily usage.
the wireless "performance" is ridiculous in master.
not only for you!
I wonder why some people seems to be happy with it, then. For me master is mostly good, except that it seems impossible to use multicast without flooding the wired network with the data.
I wonder why some people seems to be happy with it
Because they only use wireless for WhatsApp and browsing the www from the tablet.
In master, If I do 1 or 2 large file transfers between my NAS and laptop it will crash the 5Ghz band, regular usage doesn't usually cause issues. I suspect it has to do with buffers like hnyman mentioned earlier. It's been this way for a while in terms of master builds.
@dissent1, @nbd, @blogic, @gcobb or anyone else who knows their way in the switch driver, I need some assistance. On the R7800, whenever IGMP snooping is enabled on the switch, all IGMP queries (but not reports) ends up at eth0 (port 0 on the switch chip). This renders multicasting useless, since the queries from the server (which is not in WAN) never reach the clients (which are also not on eth0/WAN). I believe this has to be a bug in the switch configuration. All the details in this thread.
Hi all,
recently I became an owner of new R7800. From the beginning I was annoyed a little by bright leds lighting through the ventilation grill. When you look at router from the top it is ok, but if it on the level of your eyes it very uncomfortable.
So I decided to improve it. Instruction in the beginning of this thread was very helpful, I used a plastic card to prong all the latches.
Also I removed a cover and it revealed weird picture how they used thermopads, it was scruffily with little contact area. Not sure why it is so.
and fingerprint
It is good that I had some thermopads with 1.5mm thickness in my PC stuff so I added it to missed area
I didn't measure temperature before and after but I think this way of cooling is more proper.
Then I took isolation tapes and stick it in front of leds
Result is excellent, lights come only from the top indicators and no excessive bright lights through the ventilation grill
And then I installed LEDE Reboot SNAPSHOT r4675-e5e6045130 
I found wireless performance is good.
What is not so good, but I hope it will be sorted out
- ping to the provider on my ancient Asus RT-AC66U router was 7-8 ms, now on R7800 LEDE it is 8-10 ms
- usb3 is very slow comparing to Voxel f/w
what is the issue with wireless performance, is it LEDE related or with all open source?
Before I swtiched to LEDE I measured wire, wireless and usb performance on latest Voxel V1.0.2.33SF f/w
Then I compared it with LEDE and at least wireless is even better on LEDE.
On Voxel on 5G I had 53Mb/s upload and 64Mb/s download (speed is measured by copying of files with size 3-5 Gb)
Now on LEDE it is 64Mb/s upload and 66Mb/s download, i e nearly 520Mbit. Isn't it is good result?
On the LEDE Reboot SNAPSHOT r4675-e5e6045130 I didn't face an issue, what large files do you transfer?
I created file with 45Gb (45891Mb) and downloaded it in 11 min 52 sec with no any crashes, i e it is 64Mb/s or 515Mbit
actually only in master.
the Stable Release builds is fine.
ok, got it
Is this 520 Mbit the best speed on 5G that can be achieved or it may be even better? (on any firmware)
I have a question regarding a performance issue I am seeing, but which I don't know how to investigate.
In my setup I have a regular WAN connection for internet traffic, which provides 300/300 Mbit/s. In addition I have an IPTV connection on a different physical interface. Usually the STB will be connected directly to this interface, where it receives an IP from a DCHP server, and then uses IGMP to subscribe to the TV channels. However, I want to move the STB to my LAN instead (for good reason, but I'm trying to keep this post short(ish)), so I am using a VLAN capable switch to provide a connection from the IPTV interface to the router. This VLAN is then received at one of the switch ports on the router, and for all intents and purposes acts like a WAN dedicated for IPTV.
The STB gets an IP (in my LAN) from the router, and I've set up appropriate forwardings and routes to make this work. However, there is an issue with performance, because when I load the WAN connection with Speedtest, there is some stuttering on the TV, indicating that the multicast stream is being disturbed,
The TV multicast is probably somewhere around 5-10 Mbit/s, while the WAN is around 300 Mbit/s, and the hardware should be more than capable to handle this without any issues. Now, for the IPTV eth0 isn't involved at all, but for the internet both eth1 and eth0 are involved. The computer and the STB are both in the LAN, but the Speedtest is obviously using eth0 for internet connectivity.
If I copy some large files between a NAS and my computer (both in LAN) I see no issues, but that's (I guess) because it isn't a routed connection (the switch handles that in hardware). However, in the first test, both connections are routed, so obviously loads the CPU. How do I investigate this to find out where the bottleneck is?
I've tested irqbalance, and it does balance out the irqs quite nicely, and possibly lessens the issue somewhat, but not nearly enough.