NanoPI R2S is a great OpenWrt device

I've been looking at this device also but see that it needs an SD card yet seems to come with an openwrt os on it.

I don't need external storage, I just use the /etc/ for a few custom config files and /tmp for temporarily running some things. In my case, could I write what I need to it using openwrt and not need an SD card or is an SD card a must.

The operating system lives on the SD card and therefore is required. It does NOT come with a pre-installed/embedded OS.

1 Like

It's just not explained well on amazon. It says it comes with a version of openwrt, doesn't mention being on an SD card so one would think it means it resides in memory and the SD card is to expand or use as storage.

Thanks.

Before you purchase, what is your internet speed and do you use SQM?

_FailSafe, do you mean me?
Depends where I'm using it. At home, 50Mbps, at the office, 500Mbps.
No SQM.

Hi,

Can anyone run the tests listed in this thread?

Link to python tests:

I'm interested to see what this SOC can do.

Well I was specifically asking about the internet connection you intended to use with this device. I have had an overall negative experience when trying to shape my 480mbps/24mbps connection via SQM+CAKE. Was hoping to save you some headache depending on how fast your connection was.

1 Like

Is there any roadmap when the NanoPI R2S might be imported/integrated into the stable release branches?

1 Like

The existing ones (19.07.x/ 18.06.x), never - but it will be part of the future 20.xy branch.

1 Like

Thanks @slh! Any rough estimate when the time for 20.xy will come?
How do you feel about using snapshots for the R2S in a production environment. Basically we use NAT, iptables and basic dhcp/dns features with some glue logic around? Is there any way how I can as a non-developer understand how far a particular snapshot is developed to be close to it's prod or better stable state?

The NanoPi R2S is quite literally unbrickable, take out the sdhc card and you're back to clean sheets - so what do you have to lose?
Keep a spare, with the previous OpenWrt (snapshot-) version around, if you'd like to be able to revert to a known-good version. Any size or performance class will do, so you can use the cheapest ones from the garbage bin.

Disclaimer: I've been using using self-built master snapshots for the last >13 years, regularly refreshing my builds between once a week to at least every ~6-8 weeks on average. So far I haven't lost any hardware due to OpenWrt (admittedly, reasonable recovery mechanisms range high on my requirement list for new hardware - but I didn't need to execute those that often either). Bugs can happen, but if no one tests master snapshots in time, they would end up in the stable releases just as well (and sometimes they do) - it's just impossible to regularly test >750 devices in practice, nor just the hardware agnostic tasks regular users might throw at their devices. IMHO the master branch is remarkably reliable.

2 Likes

Thanks @slh for sharing your experience very much appreciated!

My question was not so much towards the hardware part but in regard of the kernel/software being stable and reliable running on that hw. While reading through this thread I got the impression a lot of parts are still work in progress but that might be just my judgement - that's why I ask (;
From testing sth to being "production ready" can be quite a long road. That said I do run snapshots on our test-R2S but wasn't sure / can't judge what the current state is when it comes to being "production ready"?
That said I have had the reported heat problems when stressing the device and wasn't able to reach gbit speeds yet but all that is not necessary a showstopper for prod env. I hadn't had the hardware hangs some reported yet but that would indeed be sth. I wouldn't wanna roll out in prod.

No one will be able to guarantee your use case for you, except yourself. These are pretty cheap devices, buy a few and try it out, there is not much to lose.

I think that was the main message @slh was trying to get across.

I am soon taking the plunge myself.

cheers,
/J

Just to muddy the waters for the people who think to get an r2s, look at the https://wiki.radxa.com/RockpiE as well (similar pricing better hw). There is a custom openwrt image supplied on the site. I'm currently running a x86 (i5 5200u, dual nics) but consider one of those to play around.

One of the ROCK Pi E ethernet ports seems to be just 100BASE-T, which would be rather disappointing for router uses. You can find plenty more conventional routers that can do routing at 100 MBit/s linespeed easily.

2 Likes

100Mb WAN port not a problem for many many of us...

How are you installing the OpenWrt builds onto the NanoPi R2S?

I've tried installing the OpenWrt builds posted here, including the RockChip snapshot, @jayanta525's snapshot https://github.com/jayanta525/openwrt-nanopi-r2s/releases, as well as a few others, and none of them will boot.

For flashing the xd card, I'm using the win32diskimg tool from FriendlyElec's website that I used to flash the FriendlyWrt image, which worked fine. Should I try using the Linux dd utility instead?

I've been using belena etcher without any issues, maybe give it a try.

well the secondary nic on the r2s is via USB so not perfect as well. You can take any sbc (with the same soc) with a single ethernet adapter and usb3 and achieve the same.

I'm not trying to argue, just pointing another alternative. For the rock pi e you can put external hdd on the usb3 and use as nas as well.

In both cases, you can put a 1Gb switch on LAN port for multi LAN expansion. Rock has USB 3, major expansion option, as R2S only has USB 2. Rock also has dual band g/ac Wi-Fi plus emmc socket. Other than 100Mb WAN port, Rock has way better specs.

1 Like