firmware
firmware copied to clipboard
freifunk-berlin
Possible fix for TP-Link TL-WR1043N/ND v1 instability
WiFi on my TP-Link TL-WR1043N/ND v1 gets inaccessible and needs rebooting after high load situations.
Executing
echo 0 > /sys/kernel/debug/ieee80211/phy0/ath9k/ani
upon start seems to solve the issue. I am still monitoring, but after one week of testing, everything seems to be fine.
Can you provide us with a bit of background information about the option and why it should solve the problem?
ANI is "Adaptive Noise Immunity", but I have no clue if the bug is in the chip or the driver. Fact is, the problem is gone. My router is in a usable state finally.
It was mentioned here: https://github.com/libremesh/lime-packages/issues/144 https://www.reddit.com/r/linux/comments/203sn0/this_bug_has_been_around_for_years/ https://www.gargoyle-router.com/phpbb/viewtopic.php?t=4667 https://www.walterschlag.net/?paged=16 https://tweakers.net/downloads/42463/lede-17013.html
I can confirm that the TP-Link TL-WR1043N/ND v1 will collapse under high wifi load, this is why I replaced it by another router. Should be
echo 0 > /sys/kernel/debug/ieee80211/phy0/ath9k/ani
the standard for this router?
As a sidenote: My 1043v2 was not affected by that ani change, neither positive nor negative, wifi still stopped working afters hours under high load / many users. But it is a fix I read several times about, and it seems to help under certain conditions (maybe only on that router version? I can't remember reading about that problem/fix being version specific). So I'd vote for including that change for 1043 models.
Uptime now 19 days, still no problems (v1). Before that fix I needed to reboot up to twice a day, depending on the load.
Turning off ANI seems to have decreased load and stability on one of my TL-WR1043N/ND v1. The other seems to be unaffected, but already showed low load without it. See attached graph.
This is the unaffected node, showing low load before and after turning off ANI.
Is this something we should discuss with the upstream project (openwrt)? Is there already a discussion upstream? If not can someone open a ticket?
of course there are many ticket about that problem on other projects using openwrt (libremesh, freifunk-gluon, freifunk-meshkit, battlemesh) ... and also on openwrt (subject may differ, because atm there is no tested or acceptable fix), f.i.
https://bugs.openwrt.org/index.php?do=details&task_id=863
see also: ANI, ani-noise-immunity, weak signal detection, ofdmNoiseImmunityLevel, ATH9K_ANI_OFDM_DEF_LEVEL
I can confirm that the TP-Link TL-WR1043N/ND v1 will collapse under high wifi load, this is why I replaced it by another router. Should be
echo 0 > /sys/kernel/debug/ieee80211/phy0/ath9k/ani
the standard for this router?
yes. i am running two 1043NDv1 and they do perform better and more stable with deactivated ANI run by rc.local at boot.
