• Content count

  • Joined

  • Last visited

1 Follower

About chwe

  • Rank
    Embedded member

Profile Information

  • Location

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. chwe

    A suitable board

    kernelnaming doesn't mean anything.. Change this one: and you can call your kernel: 4.17.5-super-mega-deadpool (not recommended for armbian due to kernel package script won't work properly). That your userspace terminal looks like openWRT looks like is clear (cause you use a openWRT rootfs). If your stuff is tied so close that a change doesn't make sense you might have a look what's needed in openWRT to bring up the stuff which doesn't work (e.g. HDMI). Maybe it's only a kernelconfig issue, maybe a patch is missing (for the 3.10 kernel, you find all our patches here:, for the kernelconfig you find this stuff here: Maybe you can implement *needed feature* for openWRT on the pine and send then the patches to their structure to improve openWRT (I've no idea how they organize patches/contributions but for sure they're also grateful for all improvements - openWRT is a big 'blackbox' for me in terms of update and contribution). It's not that I don't like openWRT, it's only that I don't understand their processes good enough to give you some guidance but compare kernelconfig and patches to nail down the problem should be possible when you're patient enough (for the same reason I prefer mainline over bsp kernels cause they act more predictable and documentation is often also better)..
  2. chwe

    A suitable board

    Does a 'router like' board need HDMI? Well different people, different opinions on that. Problem is when you start to mix up those things, openwrt may have some needs to for a kernel which 'we' don't care (I'm not really familiar with openWRT). How do you ensure that your device gets regular updates (kernel and userspace side) when you've a combination whit doesn't fit to both distributions? This may need additional work to keep your device updated. For me, such router/network boards should run with a (patched) mainline/vanilla kernel or at least with a kernel which isn't EOL (the only device which doesn't get regular updates in my home is the outdated iPad and the android phone, sony decided that my phone isn't worth to get updates a few months after I bought it ). Is your software so close tied to openWRT so that a switch to debian isn't possible? I don't know how the whole kernel-update procedure in openWRT world works to keep your device up to date but IMO it's worth to keep an eye on this (the last time I used a openWRT device is years ago in a time when I didn't care much about it, now my ISP is in charge but hey, I'm sure they'll do a decent job ).
  3. chwe

    A suitable board

    both should run mainline (with some limitations) so at least there, a test 'should' be possible. The only reason I don't like my Pine64 PSU idea. His board would run in a 'larger' nertwork, guess how happy a network admin will be to have an untrusted device in his network.. Since @wtarreau 'decided' to let 3.10 go EOL this doesn't look sane to rely on an EOL kernel for a new project (besides the fact that I don't like BSP kernels that much anymore). don't sound sane to me.. Where do you start to mix things? There are so many possibilities that things can go wrong and support might be a nightmare if you don't describe exactly what you did.. Wait, is it possible to use another WiFi instead the one they sell in their store? I am confused. For the RockPro64 there's a 2x2 mimo wifi (also from pine folks). which 'could' also work (same 'connector'). But it seems that nobody tested it ever, for 3.10 kernel it would need probably a back-port and mainline might have some other drawbacks (from what I know or better from what I heard from @tkaiser, the pmic doesn't work fully for charging batteries etc.). Is there a reason for openwrt? What does debian/ubuntu miss that you find in openwrt? (not meant offensive, more in a 'maybe we find something similar in debian' manner..)
  4. I claim (cause I don't have really proof) this 'different disks' you get at the moment are mostly marketing BS (as long as we don't talk about enterprise server HDDs - doesn't mean you should go for the cheapest one you can get, as for SD-cards buy your HDDs from a reliable company). This 'WD rainbow' of different 'end-user HDDs' looks more like a unicorn to me (for sure, there's a bit firmware adjustments inside so that those HDDs act slightly different).. I think it's more important that you don't threat your HDD as worse as possible (writing small junks of data with an HC1/HDD combo placed on your washing machine with 10-20 spin ups every day might be a not the best idea... ). Having a proper and tested back-up strategy is IMO more important (a heretic would argue that a one-disk setup forces you to think about this issue whereas a multidisk setup forces you to do stupid things). For me this means that data which I need 'worldwide' accessible is stored on some random cloud service (encrypted, so that the cloud-service 'should have' no clue what I store on their service, I just use them as a dump storage provider - keep in mind, most free services don't guarantee for the integrity of your data), backuped from time to time on my NAS (which is only local accessible, it would need to much efforts to keep it and my whole homenetwork save when the NAS is open). My NAS has gets backuped weekly with an external drive (connect --> backup --> disconnect). A proper back-up strategy would also involve a off-side storage (in case your house burns down etc.). I only do this for a few GBs of data I really care.. Like a squirrel encrypted in multiple copies on USB-sticks (encryption keys therefore should then also be somewhere else than your burning house ). But a backup strategy is only good when it is tested and you know it's working! Remove the disk(s) of your NAS and test if you get your data from your backups back. How much time does it need, and is everything back you thought that should be back? (e.g. in my first run, I figured out that the data is back, but all my user-settings in OMV are lost). Data integrity is then the next thing you should think about.. A back-up of corrupted data is still corrupted when you restore it... But that might be a topic for @tkaiser cause I don't feel comfortable to suggest something how you can ensure that your data is in a good shape.
  5. chwe

    A suitable board

    Should.. Did someone ever test this? I was also curious due to 2x2 mimo together with the pine64-LTS seems to be what I look for my UPS capable IoT 'master node' thingie which doesn't need that much CPU power. But the fact that pine wrote 'only for Rock' was a bit annoying.
  6. web-server for what? For 100ds of users simultaneously or just some dash-board stuff for monitoring your IoT stuff? In case the first, I've no idea.. in case the second, well my IoT stuff with web-frontend runs on el cheapo OPi Zeros (with Arm A7 and 512mb ram which is mostly not half used). GbE is something you don't get with the RPi 3+... you get a bizarre mixture between GbE and fast Ethernet which seems to make more problems that it solves. Fresh and maintained with support is somehow... Well define support, some people need more some need less. If you think the RPi3 b+ is sufficient from 'horse power' I would go for the Rock64 (I don't have one but it's on my list). The support seems to be decent (mainline & bsp), the price is good (25$ for 1GB 35$ for 2GB ram) you get proper powering via barrel plug, you get an eMMC socket for reliable OS storage and in case you need more 'fast storage' USB3 with an attached SSD should perform ok-ish (no 'real SATA' but fast enough for most purposes IMO). And you get 'real GbE' not only 'GbE over USB2 shared with all available USBs'. RK3328 gets a decent mainline support, so chances are hight that those boards are supported for a longer time... For the OS, those 80MB/s don't matter.. It's all about randomIO. I suggest you read through @tkaisers various threads where he explains this in detail. The RPi had/has its use-cases but there aren't many anymore and for web-server like stuff there are IMO better boards in the same price-range. Worth to read: Some of the informations there are outdated, but to gain some background infos it's still worth a look.
  7. chwe

    A suitable board

    Why do you need 3 USB wifis? Is it still the 2x w-lan 1x 3g/4g setup? Then you might go for the SDIO wifi which is possible with the pine64 (
  8. is most powerful a need or a wish? 1GB ram, a more or less standard slow arm.. There you go powerful board.. what does www traffic mean? use it as a desktop replacement to watch cat content in the browser? or have a small webserver for random stuff? For the first, I'm not sure if you're happy with 1GB ram, for the second hmm there's often not that much power needed.. The same here.. The RPi is for sure not one of the faster ones (e.g. when we talk about sequential speed - which IMO mostly doesn't matter, it's mostly randomIO which matters and there it's a combination of SD-Card and board). I suggest that you describe your use-case better. This would make it easier to 'find' a board which fulfills your needs.
  9. chwe

    Daily (tech related) news diet

    The article describes a weirdo app I never heard about before.. It's some strange mixture of 'social media' mixed with an easy way to pay without credit card.. In it's default every transaction is public.. Funfact, they have some sort of a json string which can be accessed by everyone with a bunch of informations. really? do people really want that their financial transactions are shared that way (not how much you spent, but conversation related to it, full name nickname and to whom you paid, including their data)? Do I miss a new hipster trend? Should I do this too? And my question.. Why should 'public' be default in such an app.. well, I'm curious what others think..
  10. chwe

    What about Google Home

    Did you make some progress on the network stuff?
  11. chwe

    Librecomputer Renegade RK3328

    you may go though this PR: never did it, never test it.. but for sure more recent than yours..
  12. chwe

    quick review Quick review of Banana Pi M2+

    So, with a proper patch it should be possible to distinguish between both board revisions and write a DT for each. I guess it 'should' be possible to do this with a DTBO (knowing that there are some parts not solved yet). But when our defaults are working for both revisions (probably with no usage of the GPIO CPU voltage regulation), I think we're on the 'good side'. Then it's up to the owners and/or boardmaker to deliver a sane patch which benefits from the new voltage regulation. When nobody of them cares, well I don't see much of a problem. I did this once for the BPI-M2-Zero cause I own the board. I don't write such patches when I don't have both revisions to test or know someone who has the other one (I need at least one, cause my DT-Fu is limited... ).
  13. chwe

    quick review Quick review of Banana Pi M2+

    IMO, keep it 'as is'. It should be a bit 'to warm' but it should be save as far as I understand? Better have a board which is a bit to warm than the support nightmare when those boards start to crash due to 'undervoltage' of the cpu. As long as nobody has the 1.2rev for testing, IMO we should go for a save approach (stability vs. performance wise).
  14. chwe

    quick review Quick review of Banana Pi M2+

    From the schematics, it looks like 1.3V will be default. Well, we will see... That the board might not be that interesting.. normal for H3 boards. According to dl page, there are still some downloads. At least what I appreciate, they inform proactive that they changed something, and also what they changed. The OPi Zero raised 8.49$ for the 256mb and 10.49$ versions (+ shipping 5.50$ here in switzerland). Depending on your needs, those boards are still good. For most of my needs the OPi PC+ is sufficient (29$ including shipping) the BPi M2+ is around 40$ with shipping, but it has GbE. So it's up to the user to decide what's more important. For this price a rock64 gets more and more interesting. Powerful and cheap.
  15. @Igor is it possible to get a valid TLS certificate for the forum too? I realized recently that has now a certificate, whereas doesn't have one. Well, that was even faster than my post... Can you read minds? Do we plan to roll out mind read as a service soon? I think re-structuration and re-re-structuration was done quite often, and I don't really see that this will help that people post their stuff in the 'right' sub-forum. If you read through this one you see that my opinion on this was slightly different in the beginnings.. turned out that I was wrong (or that I'm wrong now, who knows). Sometimes It's not really clear where a question fits perfect (e.g. it might be kernel-related or not..). Somehow, developers can 'smell' why a beginner fails and why *random issue* shows up. Is it worth to 'lose' the fast answer to be 'more polite'.. IMO (at the moment - obviously my opinions can change over time) no. When you step into 'Arm Linux' from x86 and or RPi world things differ.. There aren't that many people taking care that you don't have to care about how your thingie works, what's needed to keep it working and why *random simple task* might not be that easy as you thought. There are two red messages showing up before your account gets validated by a mod: and for sure, we might shorten both of them a little bit, but I'm not convinced that this will help to get more 'proper' first time questions. Sometimes a grumpy answer helps, sometimes it's counterproductive. I might have a 'blind spot' on this (for sure you'll find some grumpy answers I gave - everyone has bad days)...