its either that, or my routine patching of linux machines (via

apt dist-upgrade

) has also changed something

hmmmm

i'm not seeing a

cni0

bridge on any of my agents, though i do see a

flannel.1

interface.

do you know/remember if agents should have a

cni0

?

ignore that - cni0 appeared when i created some pods, and i could confirm pod-pod traffic on the same host works, but cross-host (using the overlay) doesnt

sounds like either an ip forwarding or firewall issue. Did you check firewall?

ufw

is disabled - but i didnt check iptables

i needed to setup logging for iptables anyway, maybe this is a good time to do it

cat /proc/sys/net/ipv4/ip_forward

is

1

on all nodes

another interesting observation with the pingtest. on each linux host, i run

tcpdump -vv -nn -XX -i flannel.1

then from my laptop, i run:

kubectl exec -it -n pingtest pingtest-56694d566d-sh6dw -- ping -p 33 10.42.1.6

i see continuous ICMP echo requests on the two hosts.

Interesting…could this possibly be an MTU issue?

found my issue: https://github.com/k3s-io/k3s/issues/565

specifically: https://github.com/k3s-io/k3s/issues/565#issuecomment-504766302

not sure why it happens though since i did several fresh installs and multiple restarts and reboots on different versions of k3s

will have a look tomorrow

Strange: Debian 11 Container (LXC) is not showing $OSTYPE via netclient-install.sh So it says: "This OS is not currently supported via automated install" Okay, ubuntu container same thing uname -i would work root@UbuntuTest1:~# uname -i x86_64 so if $OSTYPE is not working you should still check uname -i

you must have tested this within minutes of merging to master!!!

just updated the script lol

it just checks "uname" now

LOL, thx

I have a weird issue with one of my clients on an oracle vps. After some time it cannot checkin although it is still connected to the network. Get a

trouble decoding file

error. Have left and rejoined the network a couple of times and the problem always comes back. No issue with another client on a different oracle vps

Part of me thinks we should just ignore Oracle altogether…

Which file is it unable to decode?

Not sure

You get a different error on pull or push i will check later

mkasun@oracle ~> sudo netclient pull -n netmaker trouble decoding file panic: runtime error: invalid memory address or nil pointer dereference [signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x9ecacc] goroutine 1 [running]: github.com/gravitl/netmaker/netclient/functions.Pull(0x7ffc359c1ebf, 0x8, 0x1, 0x0, 0x0, 0x0) /root/netmaker/netclient/functions/checkin.go:157 +0x6c github.com/gravitl/netmaker/netclient/command.Pull(0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, ...) /root/netmaker/netclient/command/commands.go:184 +0x6a main.main.func5(0xc000020d00, 0x1d, 0x38) /root/netmaker/netclient/main.go:290 +0x158 github.com/urfave/cli/v2.(*Command).Run(0xc0001966c0, 0xc000020240, 0x0, 0x0) /root/go/pkg/mod/github.com/urfave/cli/v2@v2.3.0/command.go:163 +0x4dd github.com/urfave/cli/v2.(*App).RunContext(0xc0002fe000, 0xcce8d0, 0xc0000b4010, 0xc0000bc000, 0x4, 0x4, 0x0, 0x0) /root/go/pkg/mod/github.com/urfave/cli/v2@v2.3.0/app.go:313 +0x810 github.com/urfave/cli/v2.(*App).Run(...) /root/go/pkg/mod/github.com/urfave/cli/v2@v2.3.0/app.go:224 main.main() /root/netmaker/netclient/main.go:371 +0x24b1 mkasun@oracle ~ [2]>

mkasun@oracle ~ [2]> sudo netclient push -n netmaker trouble decoding file 2021/10/20 20:34:26 [netclient] completed pushing network configs to remote server 2021/10/20 20:34:26 [netclient] success 2021/10/20 20:34:26 yaml: line 58: did not find expected key mkasun@oracle ~ [1]>

mkasun@oracle ~> sudo netclient leave -n netmaker trouble decoding file 2021/10/20 20:35:30 [netclient] error attempting to leave network netmaker 2021/10/20 20:35:30 yaml: line 58: did not find expected key

can you cat the netconfig file?

maybe it got corrupted