There are some other config and data files that may reference old network interface names that you need to adjust after swconfig->DSA conversion. At least /etc/config/luci_statistics, the RRD data files under /var/rrd, and /etc/avahi/avahi-daemon.conf
Hey guys, I want to know whether is NanoPi R4SE equal to R4S Enterprise Edition and does OpenWRT 24.10.0 support R4SE?
Because on the https://openwrt.org/toh/friendlyarm/nanopi_r4s_v1 wiki it mentioned
"...and the R4SE with onboard 32GB eMMC are not officially supported however have custom builds available. " but under supported version I can see support for R4S Enterprise Edition. Thanks
Upgraded bthh5 from 23.05 using the luci sysupgrade. No problems, retained config.
Still intermittent crashes and reboot on this build. 3 in last 24 hours. Using relayd.
R4S Enterprise Edition supports the R4S with EEPROM (mac address).
I run the R4S without the eMMC (normal SD).
From what I understand, the driver for the eMMC is closed source, so no official (open source) build from OpenWrt is possible.
FriendlyARM provides their own OpenWrt build, including eMMC driver.
Thats possibly what they mean by a "custom build".
However, I seem to remember that its possible to use a regular SD card w the eMMC model, with the vanilla (OS) OpenWrt version. Maybe someone here can comment further.
at firmware-selector.openwrt.org, if you type in R4SE.....
The more I look at my upgrade options vs other's experiences, the more confused I get.
Others have apparently successfully upgraded 23.05 -> 24.10 via Attended Sysupgrade.
If I look for upgrades via Attended Sysupgrade, I get the following status---
"The device runs the latest firmware version 23.05.5 - r24106-10cc5fcd00"
However I'm running an R4S Enterprise (MAC address) anyway, but do NOT have the Enterprise version installed, so quite possibly, I cannot just do an in-place upgrade.
So, I need to understand the Firmware Downloader better.
As in, whats the method for installing the latest FM image AND retaining the settings & extra modules?
To see firmware versions in the next series in Attended SysUpgrade, click on the "Configuration" tab, click the check box for "Advanced Mode", then select the "Save & Apply" button. Once you see confirmation, go back to the "Overview" tab and the newer versions should show up when you "Search for firmware upgrade".
Here is what I do on my non-enterprise/non-emmc R4S using a new/different sd-card so I can easily switch back to the previous working setup in the short term:
-
On R4S - Get the list of installed packages from Attended SysUpgrade -> Overview -> "Search for firmware upgrade", selecting the full list, copying to an editor window on my pc, convert newlines to spaces and save it in a text file on my pc.
-
On R4S - Create a backup of the current config including my custom back-up list added in "Backup / Flash Firmware->Configuration".
-
On pc - Open a copy of the backup archive and verify "everything" is there. Do not remove original compressed file.
-
On pc - Use the firmware selector to get a new image with the list of packages from the previous step. Correct any package problems as some have changed as noted in the release notes. Review the output messages for errors or other items of interest and consider saving a copy of both on your pc. Right click on the dowload button for the image of choice (ext4 or squash) and "save image as" to your pc. Save the sha256sums.
-
On pc - Check the sha256sum of the downloaded file and unzip the downloaded file, ignoring the "trailing garbage" notice, then flash the new blank sd-card with the unzipped image.
-
On R4S - Enter
poweroffon the R4s ssh command line and unplug the power cable and remove the sd-card, insert the new sd-card and apply power. Connect my pc directly to the lan port and log in to the browser at the default address of192.168.1.1using the default usernamerootwith a blank password. Select System -> Backup / Flash Firmware -> Actions and "Restore backup" checking for confirmation. Reviewdmesgandlogreadand save their output for future reference. Consider doing the same running the previous version on the older sd-card on a fresh boot before switching to the new OpenWrt version.
Reboot the R4S, put the normal LAN cable back in the LAN port and it should be running normally on the new image. Test thoroughly and grab boot logs again.
Note that you might need to clear your browser cache if luci seems to show bad info or is flaky. I usually reboot again in case there is any weirdness with applying all the config in a running state. I've seen major and minor issues in the past but all is well when booting into the new config.
Alternatively, on your pc, you could duplicate the data on your current sd-card to a new one. On the R4S - verify it boots and runs and then SysUpgrade with one sd-card keeping software and settings. You still have one copy of v23.x on the other sd-card to revert to if something is wrong with the new installation.
3 Likes
I appreciate the detail given. I do have image copies of my current OpenWrt install, inc an extra SD w up-to-date OS install.
Thanks for explaining how to use Sysupgrade. Adv mode really helped.
I'm thinking in my case, I would like to upgrade the OS to Enterprise version, since its a closer match to my HW. I've already added in the Wireguard modules into the FW Selector & downloaded the resulting FW to PC.
I also have a settings backup. I guess I can play around with a FW install & settings restore, since I have a backup SD anyway.
1 Like
I used my process in January to upgrade from 22.x to 23.05 and then a couple weeks later, from 23.05 to 24.10 and found zero issues.
I think the difference between plain R4S and Enterprise is minimal. Scan your config for "mac" address etc and be prepared to edit, comment out, or remove any lines that apply a config defined mac in place of your burned-in-address (BIA) if you prefer the BIA one.
1 Like
Hi All,
Is there a quick hack at the command line I can do to get around issue #18086?
I fear that this is going to be viewed as so trivial the dev's won't "get a round tuit" to fix it for some time.
I probably should clarify - quick hack does not mean mucking around with /etc/config/network and /etc/config/wireless, and most of all, something that would allow me to flash a C-75 that's in a 60 foot ceiling without having to rent a ladder truck in to take the AP down, flash it, screw with the config, and put it back up.
Really simple mistakes made can cost people a LOT of money. Ever rent a ladder truck? Fortunately, I test builds......
Hi all
Been running OpenWrt on a PC Engines APU2E4 without any issues for the past five years. The update to 24.10 introduced two issues on the x86/64 machine:
-
Some changes must have been made to the 6.6 kernel compile as it now runs core 0 close to max frequency of 1GHz most of the time and the other three cores have wild fluctuations in frequencies. It doesn't seem to achieve the lowest frequency at all and core frequencies seem out of sync. Kernel 5.15 worked in 23.05 works as expected with most of the time the CPU staying at the lower frequency of 600MHz and when ramping up all four cores are used similarly; and
-
I use Yamon to track data usage. Under v24.10, the gstatic js page doesn't load and Yamon reports page "breaks". This is confirmed with the luci theme Argon not showing a bing 'picture of the day' in the login page. Defaults to the offline image rather than fetching the bing page. Two router-based webpages that don't seem to be able to fetch data in 24.10 but no issues in 23.05.
Tried with a few Snapshots and the same issues are replicated. Kernel 6.6 behaves in the same way. Tried a TinyLinux version with 6.6 kernel and CPU can achieve its lowest frequencies so it seems OpenWrt-related.
Had to revert to 23.05 to prevent the CPU from running harder, having greater power consumption and increased heat.
Kernel 6.6
Kernel 5.15
2 Likes
OK, I've upgraded to 24.10 using the R4S Enterprise Image.... so far without any problems whatsoever.... I've tried connecting to WG w one device. Worked as normal.
My procedure, (using some ideas from the help in this thread),
I copied the modules listed under sysupgrade, copied module names into the FW Selector tool, generated 24.10 image that way (1 module threw an error, deleted module from listing).
I have several backups of my settings anyway.
Uploaded FW in Flash New Firmware page.
It complained about me using the incorrect FW (regular R4S vs Enterprise).
Forced continue.
Flashed without error, then was auto-presented w Login Page.
Logged in w previous password.
Everything looks normal / little new.
Wireguard briefly tested with phone, connected w/o error as per usual.
I'm very pleased to have got this far WITHOUT ANY PROBLEMS.
My turn to CONGRATULATE DEVS.
1 Like
Just reporting that I flashed x3 Extreme Networks AP3825i AP's using attended sysupgrade without issue. However, there is a report that new devices cannot be initially flashed using 24.10.0 as mentioned here.
I also flashed an Itus Shield firewall appliance (octeon) using attended sysupgrade. The upgrade worked without issue but I did notice a LUCi bug for DNSMasq as noted here which would affect all devices.
I also noticed the DNSMasq /tmp/dnsmasq.d folder changed to /tmp/dnsmasq.cfg01411c.d so it broke my blocklist scripts. A breaking change if you use that.
Thank you to everyone involved in this!
I'm running 24.10.0 on my PC Engines APU4D4 and don't see any issues of that sort since upgrading. Have you tried upgrading the Coreboot to the latest by any chance? Here's my CPU frequency 24 hour graph with the only outlier being a 4 am daily cron job to reload banip.
Further to the above, I had to change the packet steering settings on the APU4D4 under 24.10.0 to get the same performance as I did under 23.05 where I also had it enabled. The upgrade seemed to revert those to the default disabled setting. irqbalance was installed on both versions so I can't remark on if it improved or worsened things.
1 Like
Do you know the difference between:
- Packet Steering
Enabledand - Packet Steering
Enabled (All CPUs)?
Wiki is outdated, since it is still showing it as a boolean config (yes/no), and now starting with 24.10.0 it is a numeric setting that per LuCI options it can be:
packet_steering='0'for "Disabled"packet_steering='1'for "Enabled"packet_steering='2'for "Enabled (All CPUs)"
It seems that this commit introduced this change, but I was not able to find some reliable documentation about it (other than some few mentions in the forum).
2 Likes
Thanks for providing very useful data. Our CPUs are the same (Jaguar GX-412TC cores) and should behave similarly.
I'm running the latest Coreboot before the Dasharo paywall, v4.19.0.1. I've tried multiple Snapshots from different dates and kernel versions, same behaviour. Tried a vanilla 24.10 image without any configs as well.
I'll have a look at the irqbalance and packet steering settings. It's quite intriguing that mine is somehow affected and yours isn't.
I have PCIe power management disabled and core boost enabled in BIOS.
1 Like
I don't have any specific numbers but I know that on 24.10 packet steering Enabled resulted in noticeably lower throughput than Enabled (All CPUs) on my 800 Mbps HFC/PPPoE connection. The difference was I guess roughly what I found on 23.05 between it being enabled or entirely disabled.
Updated to include that I do not have any form of firewall offloading enabled (i.e. x86/64 doesn't support it in hardware, and I found no difference with software offloading off or on).
1 Like
I'm also running Coreboot v4.19.0.1 on my APU4D4, but have not changed any BIOS settings so am assuming they're at their defaults.
Good luck!
1 Like
My router is a NanoPi R6S with a big.LITTLE architecture (4xA76 big cores, 4xA55 little cores).
I am wondering that packet_steering='1' (Enabled) should use only the big cores (A76) for improved performance, while packet_steering='2' (Enabled (All CPUs)) would also use little cores (A55) that could potentially hinder performance. But I have not found this documented so far, so just guessing.
For the R6S there is a pending PR that has a commit that improves the default CPU afinitiy configuration (following what was done for the NanoPi R4S).
However for me it is still not clear how these three configurations play (or don't play) together, specially for the big.LITTLE architecture:
- Packet Steering (Enabled x Enabled All CPUs)
- CPU affinity (configured by default in
/etc/hotplug.d/net/40-net-smp-affinity) - IRQ balancing (installing package
irqbalancewhich also configures CPU affinity?)
I don't know which CPU you are running on the APU4D4, but x64 CPUs also have big.LITTLE CPUs models for a while that would also be impacted by the configurations above.
2 Likes
I had performance issues in the past that were related to the PCIe power management being enabled. This was a while back so not sure it's still relevant. I can saturate the 1Gbps Ethernet wan without breaking a sweat.
I'm honestly stumped. Compared the htop showing kernel threads and there are hardly any differences.
I can only think of hardware differences. Do you run wifi as well? I've a WLE600vx radio installed that has never been an issue.
Htop Linux 6.6
Htop Linux 5.15
1 Like