Making an SSH connection over wifi to a RPi3 running (as of right now the latest commit) postmarketOS is unbelievably slow. I'm not sure what we can do about this and if I'm the only one or not, but right now it's unbearable. Response to typing in a character can be either instant or take a second or two.
▶ ping 192.168.1.180PING 192.168.1.180 (192.168.1.180) 56(84) bytes of data.64 bytes from 192.168.1.180: icmp_seq=1 ttl=64 time=5.49 ms64 bytes from 192.168.1.180: icmp_seq=2 ttl=64 time=2.75 ms64 bytes from 192.168.1.180: icmp_seq=6 ttl=64 time=5.86 ms64 bytes from 192.168.1.180: icmp_seq=7 ttl=64 time=4.19 ms64 bytes from 192.168.1.180: icmp_seq=11 ttl=64 time=8.67 ms64 bytes from 192.168.1.180: icmp_seq=12 ttl=64 time=4.18 ms^C--- 192.168.1.180 ping statistics ---12 packets transmitted, 6 received, 50% packet loss, time 11164msrtt min/avg/max/mdev = 2.755/5.195/8.676/1.854 ms
As you can see there is quite a bit of packet loss as well.
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information
Child items ...
Show closed items
Linked items 0
Link issues together to show that they're related.
Learn more.
@drebrez you have made the original raspberry-pi package, and @yangxuan8282 you have added wifi and Bluetooth functionality. Can any of you reproduce this issue?
@PureTryOut
the PR I created was related to RPi 3B+ wifi, have nothing to do with RPi 3B wifi
I remember used to got same issues on 3B, many packet loss when use wifi connection.
But my current pmos installation not have this issues, the kernel is from this PR https://github.com/postmarketOS/pmbootstrap/pull/1543
ping 192.168.2.106PING 192.168.2.106 (192.168.2.106) 56(84) bytes of data.64 bytes from 192.168.2.106: icmp_seq=1 ttl=64 time=7.40 ms64 bytes from 192.168.2.106: icmp_seq=2 ttl=64 time=6.00 ms64 bytes from 192.168.2.106: icmp_seq=3 ttl=64 time=5.69 ms64 bytes from 192.168.2.106: icmp_seq=4 ttl=64 time=6.47 ms64 bytes from 192.168.2.106: icmp_seq=5 ttl=64 time=7.44 ms64 bytes from 192.168.2.106: icmp_seq=6 ttl=64 time=8.45 ms64 bytes from 192.168.2.106: icmp_seq=7 ttl=64 time=7.49 ms64 bytes from 192.168.2.106: icmp_seq=8 ttl=64 time=5.95 ms64 bytes from 192.168.2.106: icmp_seq=9 ttl=64 time=6.18 ms64 bytes from 192.168.2.106: icmp_seq=10 ttl=64 time=21.6 ms^C--- 192.168.2.106 ping statistics ---10 packets transmitted, 10 received, 0% packet loss, time 9007msrtt min/avg/max/mdev = 5.696/8.274/21.631/4.531 ms
the last one take 21.6ms, and I did more test, sometimes it take long time, but at least no packet loss
maybe you can re-download the latest firmware ( should put them in /lib/firmware/brcm/ ), and check the dmesg output, if that not work, maybe you can try the latest 4.14 kernel from https://github.com/postmarketOS/pmbootstrap/pull/1543
Well, re-downloading the latest firmware from the links you posted and putting them in the proper directory seems to have caused the RPi to not boot any more. Either that, or wlan was down entirely. I've also tried the kernel from your PR, but that didn't resolve anything either.
@PureTryOut
hi, I don't know if you have solve the wifi problems, but I'm pretty sure the firmware is the correct one, it's from raspberry pi official, and most distros use it, like raspbian and archlinux arm, the upstream linux-firmware-brcm( the subpackage of linux-firmware, which is also the dependency of linux-rpi ) also use it as sources
but in case you still got issues, here is my local rpi2 pmos image build by pmbootstrap, the kernel is 4.14.44
It seems my downloaded firmware was corrupt. Connected it via ethernet, placed the firmware in correct place, and it connected again. However, still insane packet loss. It already runs a kernel compiled from your PR, so that didn't really help either...
I must say, ethernet isn't the greatest connection either, although the delays are way less. Maybe it's a general networking issue? I'd really like someone else with a RPi3 to confirm this issue.
Didn't resolve anything sadly, it's still insanely slow. I can now confirm it's also really bad on ethernet, just a little bit less. So I guess it has something to do with the entire network stack rather than the specific interfaces.
@PureTryOut: maybe someone on #alpine-devel has an idea, they were discussing the raspberry pi today.
@yangxuan8282: consider joining there as well for better raspberry pi related coordination with Alpine.
I have a badly working wifi in asus flo too. Either it is the drivers fault or it is a bad wpa supplicant config. udhcpd does not get an ip as dhcpcd starts with networkmanager....
PING 8.8.8.8 (8.8.8.8): 56 data bytes
64 bytes from 8.8.8.8: seq=2 ttl=42 time=1021.732 ms
64 bytes from 8.8.8.8: seq=3 ttl=42 time=635.403 ms
64 bytes from 8.8.8.8: seq=4 ttl=42 time=249.332 ms
64 bytes from 8.8.8.8: seq=5 ttl=42 time=1092.097 ms
64 bytes from 8.8.8.8: seq=6 ttl=42 time=704.731 ms
64 bytes from 8.8.8.8: seq=7 ttl=42 time=318.946 ms
64 bytes from 8.8.8.8: seq=8 ttl=42 time=751.767 ms
64 bytes from 8.8.8.8: seq=9 ttl=42 time=365.776 ms
64 bytes from 8.8.8.8: seq=10 ttl=42 time=1208.758 ms
64 bytes from 8.8.8.8: seq=11 ttl=42 time=208.455 ms
^C
--- 8.8.8.8 ping statistics ---
12 packets transmitted, 10 packets received, 16% packet loss
round-trip min/avg/max = 208.455/655.699/1208.758 ms
after tested on several devices(rpi 3b/3b+, xps13 9343), test include iperf and sftp speed, I think there is serious network issues on alpine (use networkmanager and current config), almost every interfaces on every devices is incredible slow
for example on rpi 3b+, with raspbian use eth0 the sftp speed is about 15MB/s,
and alpine it's 6MB/s
Thank you very much for these valuable data points! I guess the best way to go from here is comparing the NetworkManager config, and if that does not help, comparing the version numbers of NetworkManager, the kernel etc. and narrow down the cause that way.
I'm currently testing pmOS on a rpi3b again. This issue seems to be caused by networkmanager and IPv6. On my device the log gets flooded by nm messages, nm seems to reconnect constantly:
Sep 27 13:01:35 rpi daemon.info NetworkManager[1053]: <info> [1538049695.9884] dhcp6 (eth0): activation: beginning transaction (timeout in 45 seconds)Sep 27 13:01:35 rpi daemon.warn NetworkManager[1053]: <warn> [1538049695.9885] dhcp6 (eth0): the dhcpcd backend does not support IPv6Sep 27 13:01:35 rpi daemon.info NetworkManager[1053]: <info> [1538049695.9886] device (eth0): state change: activated -> failed (reason 'dhcp-start-failed', sys-iface-state: 'managed')Sep 27 13:01:35 rpi daemon.info NetworkManager[1053]: <info> [1538049695.9903] manager: NetworkManager state is now DISCONNECTEDSep 27 13:01:36 rpi daemon.warn NetworkManager[1053]: <warn> [1538049696.0063] device (eth0): Activation: failed for connection 'Wired connection 1'
Forum threads for other distros suggest to set dhcp=internal in /etc/NetworkManager/NetworkManager.conf, but that didn't work for me.
Only after setting ipv6.disable=1 in /boot/cmdline.txt my connection stays perfectly stable.
I just realized that /etc/NetworkManager/conf.d/postmarketos.conf also sets dhcp=dhcpd, which seems to overwrite the setting of NetworkManager.conf.
So I left NetworkManager.conf empty, re-enabled IPv6 and tested dhcp=internal and dhcp=dhclient (after installing it) in postmarketos.conf.
Both seem to work fine, so this seems to be an issue with dhcpd. I don't know why, because it obviously supports DHCPv6.
Maybe others can test to set dhcp=internal in /etc/NetworkManager/conf.d/postmarketos.conf and see if the networking problems get better that way. If so, I suggest to make it the default, at least on Raspberry Pis.