Probably the 22.03.6 thread, vs this 23.05.2 thread. I thought the same for a moment before I double checked and realised...
Since i installed 23.05.2 on my r7800 the irqbalance package is broken. I have the following errors in the system log:
Thu Dec 7 01:15:21 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 44 affinity: Invalid argument
Thu Dec 7 01:15:21 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 54 affinity: Invalid argument
Thu Dec 7 01:15:21 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 46 affinity: Invalid argument
Thu Dec 7 01:15:21 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 53 affinity: Invalid argument
Thu Dec 7 01:15:21 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 48 affinity: Invalid argument
Thu Dec 7 01:15:21 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 49 affinity: Invalid argument
Thu Dec 7 01:15:21 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 50 affinity: Invalid argument
Thu Dec 7 01:15:31 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 44 affinity: Invalid argument
Thu Dec 7 01:15:31 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 54 affinity: Invalid argument
Thu Dec 7 01:15:31 2023 daemon.warn /usr/sbin/irqbalance: Cannot change IRQ 46 affinity: Invalid argument
Do sombody else has the problem?
Maybe the irq channels changed?
cat /proc/interrupts
Updated my Cudy WR1300 from 22.03.5 to 23.05.2, keeping my settings. No issues.
Not (likely) related to 23.05.2 by itself, but to the irqbalance 1.9.3 version bump that changed the handling/reporting of some errors by irqbalance.
I tried to fix it with https://github.com/openwrt/packages/pull/22851
More discussion in Irqbalance throwing errors on some routers after latest build following git pull
1 Like
I flashed 23.05.2 on my ZyXEL P2812HNU-F1, lantiq xrx200. After restarting the wifi, I saw this in logread:
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.029156] br-lan: port 4(phy0-ap0) entered blocking state
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.033328] br-lan: port 4(phy0-ap0) entered disabled state
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.039528] device phy0-ap0 entered promiscuous mode
Sun Dec 10 16:20:16 2023 daemon.notice hostapd: phy0-ap0: interface state UNINITIALIZED->COUNTRY_UPDATE
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.045490] gswip 1e108000.switch: port 0 failed to add fc:f5:28:93:b5:d0 vid 0 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.054562] gswip 1e108000.switch: port 2 failed to add fc:f5:28:93:b5:d0 vid 0 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.063182] gswip 1e108000.switch: port 4 failed to add fc:f5:28:93:b5:d0 vid 0 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.071473] gswip 1e108000.switch: port 0 failed to add fc:f5:28:93:b5:d0 vid 1 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.080321] gswip 1e108000.switch: port 2 failed to add fc:f5:28:93:b5:d0 vid 1 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.087551] device phy0-ap0 left promiscuous mode
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.092199] br-lan: port 4(phy0-ap0) entered disabled state
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.128328] gswip 1e108000.switch: port 4 failed to add fc:f5:28:93:b5:d0 vid 1 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.135765] gswip 1e108000.switch: port 0 failed to delete fc:f5:28:93:b5:d0 vid 1 from fdb: -2
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.145184] gswip 1e108000.switch: port 2 failed to delete fc:f5:28:93:b5:d0 vid 1 from fdb: -2
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.153008] br-lan: port 4(phy0-ap0) entered blocking state
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.158388] br-lan: port 4(phy0-ap0) entered disabled state
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.164377] device phy0-ap0 entered promiscuous mode
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.171051] gswip 1e108000.switch: port 4 failed to delete fc:f5:28:93:b5:d0 vid 1 from fdb: -2
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.179712] gswip 1e108000.switch: port 0 failed to delete fc:f5:28:93:b5:d0 vid 0 from fdb: -2
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.187971] gswip 1e108000.switch: port 2 failed to delete fc:f5:28:93:b5:d0 vid 0 from fdb: -2
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.198514] gswip 1e108000.switch: port 4 failed to delete fc:f5:28:93:b5:d0 vid 0 from fdb: -2
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.205886] gswip 1e108000.switch: port 0 failed to add fc:f5:28:93:b5:d0 vid 0 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.216221] gswip 1e108000.switch: port 2 failed to add fc:f5:28:93:b5:d0 vid 0 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.224238] gswip 1e108000.switch: port 4 failed to add fc:f5:28:93:b5:d0 vid 0 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.233105] gswip 1e108000.switch: port 0 failed to add fc:f5:28:93:b5:d0 vid 1 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.241105] gswip 1e108000.switch: port 2 failed to add fc:f5:28:93:b5:d0 vid 1 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.err kernel: [ 408.249823] gswip 1e108000.switch: port 4 failed to add fc:f5:28:93:b5:d0 vid 1 to fdb: -22
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.315289] IPv6: ADDRCONF(NETDEV_CHANGE): phy0-ap0: link becomes ready
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.321086] br-lan: port 4(phy0-ap0) entered blocking state
Sun Dec 10 16:20:16 2023 kern.info kernel: [ 408.326062] br-lan: port 4(phy0-ap0) entered forwarding state
Sun Dec 10 16:20:16 2023 daemon.notice netifd: Network device 'phy0-ap0' link is up
I know 23.05 was delayed for xrx200 devices due to gswip problems. Should I be worried?
Same gswip messages with my TD-W8970, but I can't see anything going wrong... 
And those messages are mentioned as a known issue in the 23.05.2 releases notes (first message of this thread), as you hopefully already know.
Yes I do... + nothing going wrong = I'm not worried. At least until someone specifies I should and why...
Yes I saw that. And reading that github thread it's not clear to me if the error messages are actually errors, or noise. Nor is it clear what the impact of the errors are, if it are errors.
My WAN is not on the switch, it's DSL, so I suppose switch errors can't hit me hard.
25 day uptime on my WRT32X so far great release. I've been running various speedtests with my 500Mbit cable modem (since I'm moving over to the new GL-MT6000 soon). Thought I'd share.
WRT32X bufferbloat tests - 500Mbit / 35Mbit cable modem, some overprovisioning by the ISP, so I throttled it down a bit. Results: 494Mbits, +0ms, +0ms, A+ bufferbloat. CPU0: 92% load, CPU1: 77% load. Looks like this aging target would cap out a little over 500Mbits with SQM Cake. I'll add the results to the wiki.
[https://www.waveform.com/tools/bufferbloat?test-id=e101a8fc-f017-4eef-8f90-b27bcb783d62]
Anyway, thanks again devs!
What cable modem are you using just curious?
1 Like
Using an ~8 year old SB6183.
Without SQM it's about +20ms on dl, +40ms on ul under max load. With SQM +0/+0.
It maxes out with my ISP right at about 525Mbits. Been considering replacing it with an MB8611 since my ISP now provides 1Gbit at the same price. Just seems pointless to have any more bandwidth than this so haven't wanted to spend the $200 usd on it, but it would be nice to get just for the performance testing alone 
Slightly off topic to OpenWrt, but I recommend against buying "Motorola" branded DOCSIS modems.
The manufacturer that makes them "Minim" (formerly Zoom Telephonics) just licenses the "Motorola" brand name from Motorola Mobility (Lenovo). Minim is almost bankrupt and all the engineering staff have been laid off. It's only a matter of time before cable ISPs stop activating Motorola branded modems.
The real Motorola modem division went to Arris ("Surfboard" brand name) which is currently part of Commscope and soon going to be part of Vantiva (formerly Technicolor).
I suppose you are in the USA. If you have Comcast Xfinity, I suggest going for Xfinity XB8-T (rental, supports mid-split), XB7-T (rental, supports mid-split) or Arris S33 (does not YET support mid-split). If you have Charter Spectrum, I believe Spectrum provides a standalone DOCSIS modem without any monthly rental fee. Not sure about other cable ISPs.
Yes in US using Optimum. Good to know thanks. After reading all that I might grab a cheap "renewed" DOCSIS 3.1 off Amazon (Netgear is $90, Arris is $100) just to test out the gigabit. See how far the WRT32X takes it, and onto the GL-MT6000 unboxing soon. Then call the ISP to ask for a speed and rate reduction
This is Manufacturer Refurbished.
Most of the Amazon Renewed products are Seller Refurbished. Not sure about the quality. But if you are planning to buy from Amazon, I suggest:
https://www.amazon.com/ARRIS-Surfboard-S33-RB-DOCSIS-2-5Gbps/dp/B09VTJR48F?th=1
I have Comcast Xfinity connection with Arris S33 (bought directly from Arris Surfboard store for USD $162 including taxes back in February 2021), and GL.iNet GL-MT6000 (Flint 2) (flashed with OpenWrt 23.05-SNAPSHOT currently) which I received just few days back. Previously I was using Dynalink DL-WRX36 (which I still have as a spare).
Even with low speed plan, DOCSIS 3.1 modems have some benefits compared to DOCSIS 3.0 modems, like less congestion due to OFDM (downstream) / OFDMA (upstream) channels, mid-split (5 to 85 MHz upstream instead of just 5 to 42 MHz low-split, in North America, enables more upload bandwidth), Active Queue Management (AQM) (DOCSIS-PIE) etc. I wouldn't recommend anyone buy a DOCSIS 3.0 modem in 2023 and future, irrespective of what plan they have.
1 Like
Yea was reading about that, as I mentioned my modem is 8 years old and still works great up to 500Mbits or so, and has zero bufferbloat at 500Mbits thanks to SQM. No reason to upgrade other than just to 'test' 1Gbit.
Anyway I went ahead and ordered a renewed CM1000 for $89. Mainly I just like the black exterior since it'll sit next to my black OpenWrt router.
1Gbit is still utterly pointless to me (Netflix uses 10Mbit, gaming uses 1Mbit), and plan on calling to trying and lower the speed to reduce the bill, but first I'll do some performance tests.
I have a router (ER-X) and AP (EAP 225 V3). Both are running 23.05.0 firmware.
I get notified by my own update script that there's new firmware available (23.05.2). Confused why I missed 23.05.1 (I now know it was only tagged and not released) I checked the hardware page which states only 23.05.0 are supported on both devices. Firmware selector confirms 23.05.2 are available for both devices.
Is the table of hardware just not up to date and I haven't noticed before because I normally won't check the table and just directly download from servers?
Thank you.
They said above .1 release was skipped due to an issue, so it went from .0 to .2.
.2 is the latest.
3 Likes