Panics will basically always dump on stdout - this looks like watchdog.
I can't decode that load average, but I'm guessing you're using realtek-poe and some kind of monitoring atop it?
Absolutely +1 to everyone else's recommendations to pr_info() your watchdog feeds. Let me know if you need help building a patch, I'm curious as well.
I would really like help building a patch, I'm more a scripting user than a C developer. Thanks.
I have realtek-poe with the Prometheus exporter for stats, it causes a few bursts of activity. I was probably also clicking through the web UI.
Thanks for the info.
BTW, I tried and was able to install via the Serial+TFTP method without issue. Same firmware files as the V1, so I guess it's not that much different.
Thanks again.
I saw a reboot yesterday on a recently installed (OpenWrt 24.10.1) HP 1920 JG926A (24 port PoE 370w). I'm collecting data via prometheus, and in this case, it looks like a memory leak - something in user space. I'll set an alert for 50% memory usage and see if I can log in and see what it is.
Good catch, I'll see if I have per-process memory usage stats.
Not capturing per-process but there's a clear trend on my JG928A (reboot occurs but I had not set the prometheus exporter to auto-start which is why the gap occurs)
Comparing the JG927A (48-port) with JG928A (48 port with PoE) it looks like the userland processes on the JG928A tend to leak over time but the non-PoE version sits at high RAM usage all the time anyway?
Does anyone know the status of advanced features? I have been reading the datasheet and the RTL838X chipset seems to support some neat features. I'm interested in the ACLs since that would allow for traffic filtering at the port level instead of at the router. I also think that the Cable diagnosis hardware would be useful but I'm not sure about kernel support.
We talked about port isolation in the past, but I don't think we have offload for L3 ACLs here. Do you want L2 ACLs?
The hardware can do L2/L3/L4 ACLs with match rules. (No routing) Ideally it would be nice to be able to block/allow all traffic from a MAC, IP or port number.
I'm curious how much kernel support there is. If the kernel supports it adding userspace support should be trivial.
"kernel support" would be the OpenWrt driver implementation in this case. There is some support for offloading tc rules to the switch. See
That would be a really awesome feature to have.
As far as I know the cable test support in the downstream RTL software don't use the mainline Linux PHY cable test interfaces (they provide the userspace netlink API that ethtool --cable-test and --cable-test-tdr use), so code would need to be written to support that. You could look at the implementation for other PHY drivers (e.g. Marvell PHYs).
You could consider 'conman' for this - it will run as a daemon and log serial output to file whilst also allowing you to interact with the consoles. conman is packaged in Debian and derivatives at least, and I've used it for most of this century :o). https://github.com/dun/conman
Interesting option - I have minicom -c captuerfile.log .... inside a screen session to allow me to interact and capture logs at the same time.
Or conserver which does something similar and is packaged for OpenWrt. Any device with a usb port is suitable as console server. (With IPMI you don't even need the USB port.)
Hello.
I'm playing with dlink dgs-1100-10/me A revision, based on rtl8380 (8 TP ports) + 2 combo TP/SFP via rtl8214fc companion chip. Internal rtl8380 TP ports is working, but rtl8214fc not detected at all.
Better what i have - Realtek RTL8380 SERDES rtldsa_mdio-0:18: Detected internal RTL8380 SERDES is shown.
Maybe someone have any ideas?
Is there anything wrong about the information from https://svanheule.net/switches/dgs-1100_series?
@slh: I assume these are two different devices. https://www.dlink.ru/mn/products/1/1976.
@LynxChaus: There is no DTS for this device yet lets make a deal
win/win
DTS is based on dgs-1210-10mp-F device. But i can't understand - how describe SGMI ports connected to rtl8214fc. I'm try all variants with INTERNAL_PHY()/EXTERNAL_PHY()/EXTERNAL_SFP_PHY_FULL() - nothing.
Its EXTERNAL_PHY(). As said let's handle this via PR.
I'm talk about this https://www.dlink.ru/mn/products/1/1976.html device. In https://svanheule.net/switches/dgs-1100_series shown simple "dumb" un-managed switch without RAM.
