commit number wise there are maybe 23 that really effect us... all to do with base-files, kernel, firmware or userland...
the main thing that would effect someone's choice is the userspace versions or availability... i.e. travelmate etc, or some new apps aren't there... but these are often backported too...
in thinking / taking a look at this issue... i've added an ini option (you may test it out with just eth1 next time you are on 'current' or similar 3.5.newer);
it is a one way setting... (i.e. if you turn it off you have to reboot)
# EEE_DISABLE="eth1 eth0"
equivalent manual command(s);
# ethtool --show-eee eth1
ethtool --set-eee eth1 eee off
anyone is using nft-qos... especially if you are making use of the non mac-addr-limit options...
would be good if you can pm me your uci show nft-qos
(feel free to remove half of each mac-address if present for privacy)
i'll be messing around a bit(adding) with it in newer builds so remember to add to ENABLEDSERVICES also... before next upgrade...
anyone NOT using nft-qos if you see performance degradation or any other related issues on upcoming builds please also let me know... (or if you have any reservations now)
ultimately it should do nothing... but this may take some feedback to learn to what extent we have to tone it down when off (remove modules/whole package etc. etc.)
i am still on r17443, because i wanted to catch the error again.
but i could not. guess that is good and there is not really a problem.
smooth sailing for over 4 days now. maybe it already helped that one time i replugged the eth-adapter...just a guess.
should i still try your suggested settings?
just to be clear since i still havent figured the version naming scheme...
are the 3.5.x builds newer or older than 21.02. based on openwrt-versioning?
i never used 21.02., only your 3.x.x builds since nearly a year...
i would like to stay on the build which futurewise will be more focused on from your side.
yeah... 'current' tracks master and gets the 'latest'... I was really close to running 21.02.0 myself... but due to development and similar reasons, i'll probably stick with 5.10
release builds don't have coherent 'local' build numbering... there is no relation at the moment... 3.x means master...
I wouldn't do any extra tests... likely too intermittent to pin down...
i've been aware of 'complications' for a while (since over 6 months ago when neil1 was first having firstboot restoration issues)
for simplicity... i'd like to separate 'firstboot' from typical reboot (where I assume it would startup much more predictably)... so if it does not startup on a reboot and it is enabled... i'd like to know...
info-on-fixes
there are two or three workarounds for the firstboot situation...
disable it manually before and after upgrade
Test the ntp bypass method
Add something a bit fancier... a specific 'wait and watch for dns and or restart' during the firstboot process
i'm sure there is a fairly basic solution... but as this issue is intermittent and I don't run the service, i'm reluctant to spend the several hours needed to pinpoint where things are fouling up... especially when i'm unable to simulate the exact conditions for all users...
I run https-dns-proxy. Haven't had any issues on reboot, it just comes back up. It's disabled after upgrade, but I just re-enable it like I do SQM and irqbalance, etc. If I knew how to keep those enabled after an upgrade, I'd do it but I'm not that smart!
Can someone please advise me on how to configure Guest and IOT interfaces on the RPi4 and extend it to multiple WAP's? I would like the RPi4 to handle the Guest & IOT ip's.
The LAN WiFi works, but having issue with the other 2.
If it's a bit off topic, please comment in the other post:
I may be able to help, I run an IoT VLAN to a separate ssid and block those devices talking to the “trusted” LAN. PM me if you need a hand and I can explain what I have setup when I get some time today
masqbump for anyone testing 'current'... massive reduction in overheads/jiffies...
(not that is hugely effects us, but noteworthy)
dnsmasq-full - 2.85-7 -> 2.86-2
* Handle DHCPREBIND requests in the DHCPv6 server code.
* Fix bug which caused dnsmasq to lose track of processes forked.
* Major rewrite of the DNS server and domain handling code.
* Revise resource handling for number of concurrent DNS queries.
* Improve efficiency of DNSSEC.
[snip]
Tried this but nothing enable on upgrade. I must be ding something wrong....
First part of my wrt.ini:
##################################################################################
NORCCUSTOM=1 #rc.local negates most options here
###################################################################################
# EXPERIMENTAL=1 #enabletweaks non-standard+unexpected+risk you do not need this
###################################################################################
ENABLEDSERVICES="sqm irqbalance https-dns-proxy" [fboot/upgrade] enable+start
