Nah nowhere close to that and I am not even using hw offloading.
My new router has a 16 core 2Ghz processor that doubles the performance of the previous model that had a 32 core processor.
Nah nowhere close to that and I am not even using hw offloading.
My new router has a 16 core 2Ghz processor that doubles the performance of the previous model that had a 32 core processor.
which model are you at now ?
…holy **** thats a beast
Indeed, I am trying to make sure someone can compete or at least try to keep up with the vps crowd home nodes!!
should I be worried about the kind of gear you’re going to put online with that thing ?
What kind of cycle length? I can see a big variance over a couple of minutes but I put that down to what nodes are doing.
Probably about that, I am not able to confirm it is resolved but I watched it for about 15 minutes before I had to leave and it seemed pretty stable.
I was going from ~600mbps to ~200Mbps like a fricken yo-yo.
It’s funny how we have had 18 years of talk about best servers for the SN and only 1 month and 4 days about routers and the only thing that’s relevant really today is the router.
On Linux the tc
command can shape traffic at the source and make life a little less stressful for the router. See here for a start and please share your config if you find a good one. (Reducing those crazy bandwidth swings through traffic shaping needs to be balanced against node health…)
what metric should i be looking at to see if my router is overloaded.
am running 40 nodes.
internet connection is 750 / 100 Mbps.
router is an ASUS RT-AX58U running Asuswrt-Merlin
watching the action for a moment or 2 i see internet traffic rarely gets above 3 Mbps; RAM and CPU rarely above 50%
Most important is max sessions.
More connections require more RAM and more CPU. The 3 scale together.
In one word I would say latency increase. First you need to watch NAT table size conntrak_count vs conntrak_max, and maximize the max setting while not running of RAM on your router. Then upstream bandwidth utilization - you do not want to overwhelm your ISP with traffic, and CPU% must not hit 100%, or at least very sparingly.
(Servers have become so cheap that running hundreds or thousands of nodes is not a challenge.)
Update to my data point for Ubiquiti DM Pro (4GB RAM, 4 core CPU) and 200/200Mbps ISP: 500-600 nodes port-forwarded max under current network conditions.
The CPU is the weakest point but I have UniFi Protect running too (and quality of broadband service to my family to worry about; better err on the safe side;-)
@Josh what was the limit of the CCR that you are replacing?
Limit of connections?
The CCR2004 has a soft cap of around 1 million connections but it will dynamically adjust if there is available RAM.
CPU is ~75% @ 1000ish.
When the wife starts chasing you with a kitchen knife and the kids try to set you on fire
That’s a good early indicator
I just play dumb. Internet bad? Really! Seems to be working fine for me.
You could step it up and proactively gaslight them into thinking that they’re causing the issues.
man I can’t figure out where to even start with setting up my router haha. I figured out I have to put my isp gateway into bridge mode to get what I want out of my router but when I put it in bridge mode I get no internet from the router I’m a bit confused on what to even look for on how to fix this issue?
What did you get? my new router is sadly not performing nearly as well as I expected.
That’s why they call this testing I guess.
3rd times a charm, I love the Amazon try it out and if you don’t like it send it back policy