Max IP reservation on local DHCP?

[Dc38]

Hi all, 

When trying to set IP reserves on my LAN on an XR1000 running firmware 1.0.0.52_1.0.38, any device entry over #7 wipes and replaces the current #7. Is the reserved IP table limited to 7 entries?

Furthermore, I have 13 devices that connect intermittently when in use throughout the day, but they seem to be causing lagspikes in games for other devices that are currently connected, which is why I am trying to reserve IP's to avoid the random assignment per session per device. 

Would appreciate any assistance, especially with diagnosing the ping spikes that seem to happen when the other devices connect.

 

[Netduma Liam]

The DHCP pool should be far bigger than this, around 100 IPs I believe. Are you reserving IPs within the DHCP range or outside of the range? I'd suggest reserving them from outside of the range.

With regards to the lag we should be able to resolve that through QoS, could you share your current QoS configuration? A screenshot will be fine.

[Dc38]

Hi Liam,

I'll give that a shot. As far as the QOS setup goes, the image of the settings has given me the most stable tests, but there are still substantial lag spikes in game when the devices connect.

[Netduma Liam]

What speeds do you get from your ISP? Try going to the three-line menu in Congestion Control and click 'Set Bandwidth speeds'. Then have DumaOS auto-detect your speeds, this will make QoS much more effective.

[Dc38]

On an average day it's 840/920 down and up, using the connection bench with congestion control with anything above 65% each gives me consistently high ping spikes, jitter, and bufferbloat averaging 40ms. The bench scores from these configs will usually be A+ A+ B.

Using the bench while congestion is set to 20% each results in the most stable benchmark for me; 10ms average with the outlier spike ranging from anywhere between 40ms to 100ms. These benches will result in A+ A+ A+.

Anything below 60% congestion sliders will result in A+ across the board, but the bufferbloat graph will have spikes that look like wifi pulses for idle, upload, and download. Also, the results above are with "shared excess bandwidth" all disabled, as the sharing seemed to consistently result in higher ping. 

Any idea what would cause the bufferbloat graph to pulse like that? 

The top image is sliders at 100%, B bufferbloat. The images below are, respectively: sliders at 65%, and sliders at 20%.

The bottom most image is 65% with shared excess enabled.

[Netduma Liam]

Are there any devices connected upstream of the XR1000? If you've got anything connected to the ISP modem/router, traffic from these could cause spikes that DumaOS can't control but it can certainly monitor.

[Dc38]

Within the home, it's devices to XR1000, to bridge ONT, straight to the utility pole and beyond. As far as external ISP based devices, I do not know.

[Dc38]

On a side note, i limited the dhcp pool range between 1.2 and 1.25, and per recommendation assigned reserved IP's beyond that range. I was able to add one more ip reservation for a grand total of 8, but I'm still having issues where any entry past the latest will overwrite the immediate prior entry...

[Netduma Liam]

OK thanks for letting us know. What are the devices you're reserving IPs for? Do you know if they use Private/Randomised MAC addresses?

[Dc38]

I have a couple tvs, which I have matched assigned IP's to the reservation list, and a few tablets/phones and thermostats. I'm not sure where to check for randomized MAC addresses.

[Netduma Fraser]

When you're making these changes are you always doing it through a mobile browser? If so the interface isn't optimized for that and so could be the issue, could you try doing it from a PC and see if it works better please?

[Dc38]

I have been making the changes on a PC, the mobile browser screenshots were just for the sake of convenience.

[Netduma Fraser]

Could you provide the step by step process you're using to add a new reservation please? 

I remember I specifically tested this on another XR router and you can have a lot more than you currently have

[Dc38]

I am going into the settings tab, then accessing LAN setup within the Setup sub menu. I am adding devices off of the "Add" submenu, then clicking apply. 

[Netduma Liam]

5 hours ago, Dc38 said:

I am going into the settings tab, then accessing LAN setup within the Setup sub menu. I am adding devices off of the "Add" submenu, then clicking apply. 

In Device manager, do you have a lot of connected devices that aren't in use anymore? It may be worth clearing up offline devices by going to the three-line menu and hitting 'Remove offline devices' in case these still have stored network configs on the XR1000.

[Dc38]

I'll try removing the offline devices and let you know the results.

[Dc38]

I am still limited to 8 devices for some reason...

[Netduma Fraser]

Can you detail the specific devices you have connected? There might be one specifically causing a conflict - maybe one that isn't as mainstream as the rest

[Dc38]

Thanks for all the assistance, the latest couple firmwares seem to have fixed the issue!

 

On an unrelated note, I read somewhere that the chipset works best when the total concurrent connected devices does not exceed 11, and QOS works best with 6 or less concurrent devices.

[Netduma Fraser]

Great to hear that, thanks for the update!

I've not heard that before, if you find where you read that could you provide a link? Have you personally seen any behavior that seems to confirm that theory?

[Dc38]

Hi again, sorry for the late replies again...please see link for reference

https://wikidevi.wi-cat.ru/Broadcom/SoC,

use the "find on page" feature to find the chipset used in the xr1000, BCM6750, and under the Esystems column it says 10 devices. From my personal use, if I have some video on the TV, an Ipad, 2 phones, and a laptop running as well as a couple wifi cams streaming, I will hit lag stutters in the game when those wifi cams ping on with motion detection. 

[Netduma Liam]

8 hours ago, Dc38 said:

Hi again, sorry for the late replies again...please see link for reference

https://wikidevi.wi-cat.ru/Broadcom/SoC,

use the "find on page" feature to find the chipset used in the xr1000, BCM6750, and under the Esystems column it says 10 devices. From my personal use, if I have some video on the TV, an Ipad, 2 phones, and a laptop running as well as a couple wifi cams streaming, I will hit lag stutters in the game when those wifi cams ping on with motion detection. 

That's just referring to the amount of devices that use that specific chip.

Could you share with us your Traffic Prioritisation rules, what game it is that you're playing and on what kind of device?

[Dc38]

I'm currently afk, but I'm on Gb fiber with my qos sliders are set to 50%DL 75%UL, Allocations set to even levels for applications. I'm on a desktop playing anything from Division 2 to CS and Apex.

[Netduma Fraser]

Is it set to Always or Auto-Enable? You may need to fine tune the percentage to find what works for your specific connection, you may need a higher or lower percentage

[Dc38]

I have tried it both, swapping it up for each session I play, and I'm speculating, but it sure feels either the firmware/hardware is not up to the task lol...it seems to be adding JUST enough latency that leaving the qos completely off (no antibloat no prio settings) seems to work slightly better for upload/hit reg.

Whatever way though, as soon as other devices either ping on, I hit HUGE lagspikes up the order of hundreds...i have all of my devices in a reserved IP list, with a separate DHCP pool for new/random stuff. 

On a side note, does the QOS treat wifi traffic the same as ethernet? And how should a device using GeforceNOW be prioritized? (I noticed that it seems to like bumping up the priority for "uncategorized")